Can democracy be improved with random selection?

By John Huang @ 2024-10-27T09:31 (+17)

I'm an advocate of something known as sortition. The premise is simple. Choose people at random, to serve a finite term, in some decision making capacity. Pay them to be there. Sounds ridiculous, right?  How could we possibly trust ignorant, stupid, normal people to make good decisions? What would this look like? Why would this be better than electing our officials to office?

The Benefits of Sortition

Descriptive Representation

Imagine a Congress that actually looks like America. It's filled with nurses, farmers, engineers, waitresses, teachers, accountants, pastors, soldiers, stay-at-home parents, and retirees. They're conservatives, liberals, and moderates from all parts of the country and all walks of life. 

The primary benefit of using sortition is to create a diverse and descriptively representative sample of the larger public. Sortition is the best process to create a proportionally representative legislature. Sortition ensures proportionality in terms of race, class, gender, religion, ideology, party affiliation, cognitive ability, profession, and any other attribute you can think of, using the power of uniform random selection, and stratified sampling if desired. 

In contrast, “the very logic of election leads to unrepresentativeness because those who have the time, money, connections, and profile required to run successful campaigns are likely to be, on average, wealthy, educated, and from dominant social positions” [12]. 

A Deliberating Public

Experiments with deliberative democracy have generated empirical research that “refutes many of the more pessimistic claims about the citizenry’s ability to make sound judgments…. Ordinary people are capable of high-quality deliberation, especially when deliberative processes are well-arranged: when they include the provision of balanced information, expert testimony, and oversight by a facilitator” [1]. 

Even more compelling, democratic deliberation can overcome polarization, echo chambers, and extremism by promoting the considered judgment of the people. “The communicative echo chambers that intensify cultural cognition, identity reaffirmation, and polarization do not operate in deliberative conditions, even in groups of like-minded partisans. In deliberative conditions, the group becomes less extreme” [1]. 

How Deliberation Works

A deliberating Citizens' Assembly is usually conducted with the following steps:

  1. Selection Phase: An assembly of normal citizens is constructed using statistical random sampling. For various assemblies, samples have ranged from 20 to 1000 in size. These citizens are called upon to resolve a political question. Citizens are typically compensated for their service. Amenities such as free child or elderly care are provided.
  2. Learning Phase:  Educational materials are provided to help inform the selected deliberators. This may be in the form of expert panels, Q&A sessions, interactive lectures, presentations, reading materials, etc. Following each presentation, the Assembly then breaks into small, facilitated discussion groups to further increase understanding of the learning materials.
  3. Listening Phase: Stakeholders, NGO's, and other interested members of the public are invited to testify.
  4. Deliberation Phase: Facilitated discussions are held in both large and small group format. A final decision is made through voting. 

What has the Public Decided?

In deliberative polls conducted by America in One Room [2], a representative sample of 600 Americans were chosen to deliberate together for a weekend. Researchers found that “Republicans often moved significantly towards initially Democrat positions”, and “Democrats sometimes moved just as substantially toward initially Republican positions.” 

For example, only 30% of Republicans initially supported access to voter registration online, which moved to majority support after deliberation. Republicans also moved towards support for voting rights for felons dramatically, from 35 to 58%. On the other side, only 44% of Democrats initially supported a Republican proposal to require voting jurisdictions to conduct an audit of a random sample of ballots "to ensure that the votes are accurately counted". After deliberation, Democrat support increased to 58%. 

In terms of issues like climate change, the 2021 “American in One Room: Climate and Energy” deliberative polling found a 23-point increase in support for achieving net-zero after deliberation. Californians moved 15 points in support for building new-generation nuclear plants [3]. Participants also moved 15 points in favor of a carbon pricing system [6]. These changes in policy support were achieved in only 2-4 days of deliberation. 

Time and time again, normal citizens are able to make highly informed decisions that weaker-willed politicians cannot. In a 2004 Citizens’ Assembly in Canada, the assembly nearly unanimously recommended implementing an advanced election system called “Single Transferable Vote” (that was then rejected by the ignorant public in the following referendums). In Ireland, Citizens’ Assemblies played a pivotal role in recommending the legalization of gay marriage and abortion (In contrast, their elected politicians were too afraid of special interests to make the same decision). In France, 150 French citizens formed the Citizens’ Convention for Climate. The Convention recommended radical proposals to fight against climate change (including criminalization of ecocide, aviation taxes, and expansion of high speed rail). These proposals were unfortunately significantly weakened by the elected French Parliament. 

The Achilles heel of Deliberative Democracy is, how can we scale this process? Deliberative participation of the entire public is logistically impossible. However the scaling question has already been answered with every sample drawn by lottery. Deliberative democracy can only be scaled using sortition. The entire public does not need to participate; a smaller sample is sufficient to statistically represent the public.

Resistance to Special Interests

Political philosopher Arash Abizadeh argues that "competitive elections favour candidates with social privileges and resources, and are vulnerable to manipulation and domination by powerful partial interests, they neither treat persons as political equals nor treat conflicts impartially. They produce assemblies composed of elites partial to elite interests and values. This is why until  the eighteenth century elections were primary associated with aristocracy."

"Democracy, by contrast, was primarily associated with sortition.... And because truly random selection is invulnerable to influence by powerful interests, it tends to produce a descriptively representative 'mirror' of society rather than an assembly of elites - thereby helping to treat conflicts impartially" [8]. 

Lottocratic Efficiency

Sortition is a powerful tool for making efficient democratic decisions. By selecting a smaller sample to represent the public, only a fraction of the whole is required to participate in otherwise time (and therefore cost) intensive decisions. 

Imagine a referendum of 1 million citizens. Imagine that it takes at least 1 hour for each citizen to at least understand the referendum proposal (let alone understanding the consequences and pro’s and con’s of the proposal). Assuming a wage of about $15 per hour, the social cost of this uninformed decision is about $15 million. 

In contrast imagine 500 citizens selected by lottery tasked to make a decision, using four weeks of time, or 160 hours per citizen. Let’s imagine the state compensates these citizens at the rate of $100 per hour. The cost of this informed collective decision is then $8 million. 

Sortition produces an informed 160-hour decision at the cost of $8 million, while referendum produces an uninformed 1-hour decision at the cost of $16 million. Election fares hardly any better. With the same logic, elections produce an uninformed 1-hour hiring decision, while sortition produces an informed 160-hour hiring decision. 

The Double Edged Sword of Electoral Accountability

Unfortunately, a large body of work exists illustrating the lack of capability of voters. In Democracy for Realists by Achen and Bartels, the authors suggest that voters practice blind retrospection. "Real voters often have only a vague understanding of the connections (if any) between incumbent politicians' actions and their own well-being. Even professional observers of politics often struggle to understand the consequences of government policies. Politics and policy are complex. As a result, retrospective voting is likely to produce consistently misguided patterns of electoral reward and punishment" [pp 144]. Bryan Caplan's The Myth of the Rational Voter popularized the term "rational irrationality" for the behavior of voters. Caplan argues that the marginal cost of holding an erroneous political belief is low, due to the low probability of influencing the outcome of any election. Voters instead may vote due to the psychological benefits of supporting policies that feel good. These good feelings therefore outweigh the real harm of a policy, when factored with the unlikelihood of influencing the outcome. 

As Alexander Guerrero claims [9], electoral representation can bring about responsive and good  outcomes only if the public can hold their representatives meaningfully accountable. If we find that voters do not meet this competency, then sortition could have an advantage at holding government accountable. From my understanding of the available evidence, the literature overwhelmingly suggests that voters are not able to hold elected politicians accountable except in the most dire and obvious of economic disasters - for example, when the public is experiencing a famine [9] and therefore practices retrospective voting to remove incumbents. 

Sortition relies on the regular rotation of decision makers as its primary accountability mechanism. After the former assembly is rotated out of office, the new assembly can change the decisions made by the former assembly, and potentially even punish wrongdoing of the former assembly. Because participants in sortition are better informed deliberators, they have better capacity to make correct accountability decisions compared to the ignorant voter. 

Example Sortition Models

This section will briefly review some possible models of how sortition could be implemented. 

Review Panel for Elected Officials

One way to address the politicians' lack of accountability is to use sortition as an allotted review panel to assess and penalize elected officials at more frequent intervals - for example, an annual review. "The concept is similar to a criminal jury trial: the panel hears the case for and against the official having the standard of leadership expected of them, and based on that, can commend them, declare them adequate, or dismiss and/or fine them for falling short, with the option of barring them from holding public office again" [7]. 

An Allotted Electoral College

In a more radical model, sortition can be used to completely cut out the general election. Executive and advisory leadership would be selected by an electoral college of citizens selected by lottery. Political leadership would be selected, reviewed, and held accountable using democratic deliberation.

With sortition, a fully-fledged leadership hiring process could be implemented. That means a system to review hundreds/thousands of resumes. Then a process to select dozens of candidates for interviews. A final selection process. Then like with the Review Panel, regular performance reviews.  

Sortition allows for the complete elimination of the marketing/propaganda circus that is the modern political election and campaign (including the billions of dollars needed to facilitate elections participated by  millions of people, and the billions of dollars spent in advertising), in favor of deliberative leadership selection. 

Hybrid Bicameral Sortition

Philosophers and academics such as Arash Abizadeh, John Gastil, and Erik Olin Wright advocate for a bicameral legislature where an elected chamber is paired with an assembly selected by lottery. In the typical proposal, legislation is initiated by the elected chamber and is reviewed, approved, or rejected by the allotted chamber. Abizadeh justifies the continuation of elections as a mechanism to disincentivize political violence, "on the fact that competitive elections furnish, to forces currently shut out of government, the prospect of taking political power by contesting and winning future elections, without incurring the costs of civil war" [8].  

Alex Kovner and Keith Sutherland offer an alternative bicameral legislature  [10]. In their proposal, legislation initiated from the elected chamber only requires a minority (say, only 1/6th of elected representatives) to pass for review from the allotted sortition chamber. 

Multi-Body Sortition

Terril Bouricius envisions a six-chambered decision making system, powered by sortition, designed to maximize descriptive representation and increase resistance to corruption and domination of special interests [13]. These chambers are:

Bringing Theory to Practice

It would be outlandish to convert a regime overnight to a radical new political system. Ideally, a deliberative society would be capable of experimenting with new political systems. In practice, advocates must slowly persuade change-makers and power brokers for the privilege of experimentation (hence the existence of this article). 

Unfortunately, the vast majority of experiments of modern-day sortition serve in a temporary and advisory-only capacity (similarly to how many elected representatives only served in an advisory-only capacity to kings for many centuries). Many advocates wish for greater adoption of these temporary Citizens' Assemblies. Advocates hope that as the public becomes increasingly aware of Citizens’ Assemblies, politicians would eventually be persuaded to create a permanent Citizens' Assembly. Eventually, the Assembly might gain agenda setting powers. In the far future, the Citizens' Assembly might finally obtain the right to approve or reject decisions. Some progress has followed this course in selected parts of the world. Paris implemented a permanent Citizens' Council in 2021. Belgium also adopted a permanent Citizens' Assembly, with powers to advise parliament, in 2019. 

Modern-day jury duty is an imperfect implementation of sortition with significant problems. Juries are not selected randomly; attorneys use Peremptory Challenge to remove jurors for prejudicial reasons. Jury sizes are too small to serve as a statistically representative sample and therefore create chaos in jury outcomes. 

Simon Pek advocates for the adoption of sortition in worker-owned firms in order to counter what he calls organization degeneration - “the tendency for a small oligarchy of unrepresentative workers to control democratic structures at the expense of the participation of everyday workers, … [that] occurs naturally as worker-owned-firms become larger and more complex” [11]. Other small-scale democracies that could benefit from sortition include housing cooperatives, unions, selecting board of directors for nonprofits, and homeowner associations. The US-based nonprofit Democracy Without Elections currently operates as the first lottery-selected nonprofit board in the world (full disclosure, I serve as one of its temporary rotating board members in 2024-25, and I am a frequent advocate for the organization). 

Oliver Milne also proposes the use of a dual board, similar to the proposal for an Allotted Electoral College, to run a democratically controlled corporation. Milne proposes selecting about 25 salaried jurors, by lottery, as the Citizen Board. The Citizen Board’s primary task is to select and monitor a Professional Board tasked with actual administration of the corporation.

Using Sortition within Effective Altruism

Sortition facilitates scalable, deliberative, democratic decisions. This capability could be applied to charitable giving. As Effective Altruists (EA) understand, the computation of utility maximizing decisions is no easy task. Sortition presents a method of reducing cognitive load for practitioners. Imagining 1000 EA practitioners, 50 practitioners could be selected by lottery to serve a finite term in an EA Assembly. The 50 practitioners could then be tasked with intense research, fact finding, deliberation with one another, hiring staff or advisors, to produce guidelines on what charities and issues to support. This EA Assembly thereby reduces the cognitive load of the other 950 practitioners. With regular lottocratic rotation (With perhaps about ⅓ of the EA Assembly rotated out each period), new assembly members can review the work of the past to facilitate some accountability that EA objectives are being met.

This arrangement may be more effective than an elected assembly. Elections in contrast require fact-finding and monitoring from 1000 EA practitioners to evaluate the performance of the elected assembly. EA practitioners that make monitoring sacrifices for good reasons (for example to devote more time towards productivity) may inadvertently make harmful electoral decisions. 

High Impact Political Reform

There's a lot of proposals out there about how to make a better functioning democracy. Proposed reforms have included:

None of these proposals address the elephant in the room - the problem of voter incapacity. Condorcet's Jury Theorem only works when a majority of voters are able to come to the right conclusion. As experiments in deliberation show, informed deliberation is able to cause massive changes in opinion and reverse majority opinion on a variety of issues. Informed citizen deliberators make different decisions than ignorant voters. From all examples, whether it be climate change or biodiversity, informed citizens tend to choose the greater good, when voters do not. 

Sortition is the only technology we have to democratically raise the competence of citizens. That is why I believe sortition deserves the attention of the Effective Altruist community.

A list of Sortition Advocacy Organizations

References

  1. J Dryzek et al. The Crisis of Democracy and the Science of Deliberation. Science, 2019.
  2. J Fishkin, L Diamond. Can deliberation cure our divisions about democracy? Boston Globe, August 2023.
  3. Tyson, Mendoca. The American Climate Consensus. Project Syndicate, Dec 2021.
  4. J Fishkin, A Siu, L Diamond, N Bradburn. Is Deliberation an Antidote to Extreme Partisan Polarization? Reflections on "America in One Room". American Political Science Review, 2021.
  5. Citizens' Assembly. https://participedia.net/method/citizens-assembly. Accessed 2024 Oct-19.
  6. America in One Room: Climate and Energy. Participants at T1 v T2. https://deliberation.stanford.edu/news/america-one-room-climate-and-energy. Accessed 2024 Oct 19.
  7. O Milne, T Bouricius, G Flint, A Massicot. Sortition for Radicals. Citizens' Assemblies and Beyond. International Network of Sortition Advocates, 2024.
  8. A Abizadeh. Representation, Bicameralism, Political Equality, and Sortition: Reconstituting the Second Chamber as a Randomly Selected Assembly. Perspectives on Politics, 2020.
  9. A Guerrero. Against Elections: The Lottocratic Alternative. Philosophy & Public Affairs 42, no 2, 2014.
  10. A Kovner, K Sutherland. Isegoria and Isonomia: Election by Lot and the Democratic Diarchy, 2020.
  11. S Pek, Drawing Out Democracy: The role of sortition in preventing and overcoming organizational degeneration in worker-owned firms, Journal of Management Inquiry, 2019. 
  12. T Malleson. Should Democracy work through elections or sortition? Politics & Society 2018, Vol. 46(3) 401-417. 
  13. TG Bouricious - Democracy through multi-body sortition: Athenian lessons for the modern day. Journal of Public Deliberation, 2013. 

Arturo Macias @ 2024-10-28T05:35 (+2)

I would use it for expert panels (above all, the Supreme Court), never for the executive or the Legislative.

https://forum.effectivealtruism.org/posts/PyqPr4z76Z8xGZL22/sortition

John Huang @ 2024-10-28T19:24 (+1)

In my opinion you've neglected sortition's ability to vastly improve the capacity of citizens to make informed decisions. Improved capacity makes sortition the better way to select, and hold accountable, political leadership compared to elections. 

Arturo Macias @ 2024-10-28T22:24 (+1)

But even if you think that more can be done, what do you think as its use to create “expert panels” including the Supreme Court?

It looks so natural that it is increíble that Sortition (among professional justices) for High Courts is not universal.

ChristianKleineidam @ 2024-10-31T16:25 (+3)

When choosing justices on the one hand you want to choose them meritocratically and on the other hand, you want that the views of the population are well represented.

For justices in lower courts who are just supposed to rule in the way the higher courts would, you can just pick them meritocratically without problems if the higher courts are selected in a way that represents the view of the population. 

Arturo Macias @ 2024-10-31T18:06 (+1)

I disagree. Blindness is the main attribute of Justice. Rawls and the Romans were right. Meritocracy is nice at the gates of the career, but how do you measure merit among the chief justices? Any mechanism different from the lottery will become a battlefield.

Moreover, I don’t believe in merit among the experienced Justices. Law is not like chess or Physics. It is about consensus, and intuitiveness. There is not a real object to be discovered by the jurisconsult, but a mix of system, continuity, social agreement, and a bit of game theoretical intuitions. Who is the “best” at that?

Regarding the views of the population, I am for judicial review by the legislature, but not in the late stages of the career, because the closer to the high court is the political intervention, the higher becomes the risk of capture.

ChristianKleineidam @ 2024-11-11T08:32 (+1)

When speaking about "merit" of judges, the useful for what that means in practice. 

For low-level judges, that often means exam scores. In merit-based systems, you need good exam scores to become a judge.

For experienced judges you can measure merit by how much of their judgements get overturned by higher courts. A judge who constantly makes judgements that get overturned is bad at seeking legal consensus. 

In common law jurisdiction you could also measure how often the opinion of the judge on cases get cited by opinions from other courts.

Arturo Macias @ 2024-11-11T10:07 (+1)

Yes, there are some measures, but beware of Goodhart Law: if you over-incentive consensus, you get herd behaviour. Many "consensus building" mechanisms end producing the same kind of problems as "peer review": conformity, statu quo bias, and above all, guild mentality. In Law, external measures of goodness (that counterbalance statu quo bias) are even more difficult to create than in academy...

https://www.palladiummag.com/2024/08/02/the-academic-culture-of-fraud/

John Huang @ 2024-10-29T00:03 (+1)

Sure, I think your proposal is a great idea. 

ChristianKleineidam @ 2024-10-28T21:34 (+1)

If you have a bicameral legislature, having one chamber be selected via sortition and the other be selected via a more conventional election would be one good option.