Random Acts of Democracy

Sometimes a little randomness is the key to a more effective democracy.

A kleroterion (Ancient Greek: κληρωτήριον) was a randomization device used by the Athenian polis during the period of extreme democracy to select citizens to the boule, to most state offices, to the nomothetai, and to court juries.

Sometimes a little randomness is the key to a more effective democracy.

At first blush this is a remarkable assertion. Our faith in democracy is shaken when elections dangle on the vagaries of a hanging chad. Or settled precedent is abruptly overturned by a partisan court. We are justifiably outraged when innocence depends on on zip code. Or the latest twitter meme.

As Blackstone argued in Commentaries on the Laws of England (1783), the law should be “ permanent, uniform, and universal “.

But perhaps stability can emerge out of chance. After researching over 2500 years of democratic governing experience, scholars such as Bernard Manin observe that randomness, historically speaking, was the ultimate guarantor of fair elections and predictable lawmaking.

Beginning with ancient Athens in 507 B.C.E:

For centuries despots and political elites controlled the city-state. Outraged citizens, refusing to accept perpetual chaos, implored the populist leader Kleisthenes to free them from this vicious cycle.

His innovative solution was systemic randomness.

After forcing existing cliques apart, Kleisthenes established a vibrant democracy¹ with representatives chosen by lottery. In lieu of corrupt elections and hereditary positions, volunteers (free male citizens in good standing with both god and mammon), randomly served on enormous juries of five hundred or more. They filled government bureaucracies, the general legislature and even served as chief executive for a day. Citizens could question any law, and demand accountability from any representative.

Sortition ruled. Political parties endured, but they governed Athens in random order², diffusing their authority. Independent, ordinary citizens replaced dishonest partisan bureaucrats. Large, ad hoc juries could not be bribed, and the courts handed down universally admired decisions.

Confidence emerged from chaos.

Was there an occasional “clunker”? Of course, but only for a brief moment in time. Compared to Washington DC, where pedophiles, conspiracy wing-nuts and dark money candidates serve in Congress, the luck of the draw is a refreshing alternative to the ballot box³.

And sortition wasn’t limited to ancient Greece. Venetian governments filled their legislative bodies by lottery until the end of the 18th century, and even the republican Montesquieu thought “the suffrage by lot is natural to democracy.”

Americans are unlikely to pivot to systemic sortition, but “residual randomness” continues to inform our representative democracy⁴. At the local level, juries are filled by lot. Indeed, jury service, inspired by Athens⁵, is one of our few surviving direct-democracy roles.

States are “laboratories of democracy” where messy political experimentation and citizen referenda are often the precursors of tomorrow’s national policies.

And there are many more examples (see end-notes for a detail discussion).

Given its long, successful (and mostly forgotten history), how can we adapt systemic randomness for a modern and increasingly complex world⁶?

• In government contracting, don’t favor the lowest bidder. Instead, any proposal that meets a broad range of criteria is acceptable, and the winner is chosen by lottery. Randomness opens space for ingenuity, thwarts rigged bidding, and shakes up the procurement system.
• For judicial appointments, both parties nominate candidates to enter a vetted pool in proportion to their seats in Congress. Judges are appointed from the pool at random. Sortition blocks the majority’s impulse to pack the bench, and more faithfully reflects the aggregate political will. A lottery could help restore legitimacy to a tarnished court.
• And if we did pivot towards sortition, representative elections are a prime opportunity. When the top candidates are within 5% of each other in votes, the winner should be chosen by lot! Which sounds anti-democratic, but consider the reality of elections: Pervasive distortions, like hanging chads, gerrymandering, voter manipulation, incumbency, primary capture by extreme candidates, Super PACS, wealth, decision fatigue and low voter turnout may, in aggregate, overwhelm a 5% difference in certified votes. Randomness would also moderate concerns over a few unreadable ballots deciding a tight race. And a lottery would eliminate the need for complex ranked voting schemes. Why should we presume a narrow electoral victory is fundamentally more legitimate than the blind impartiality of fate erasing these systemic democratic distortions?
• Similarly, an electoral college of 5,000 randomly selected volunteer citizens might outperform the circus that is our presidential system. Even a national popular vote can be held hostage to the biases of a captured primary system, voter suppression, dark money, Russian disinformation, or the temporal vagaries of an SNL skit.

Lotteries bend the arc of democracy towards justice.

Representative governments materialized in conjunction with a ruling-class averse to sharing power. Anxious that a vibrant democracy might compromise their property and status, they restricted the voting franchise and limited direct citizen participation. Perhaps the tensions of our flawed representative democracy will compel a new generation to seek inspiration from Kleisthenes, and once again enjoy the benefits of systemic randomness.

Important supplementary material TL?;DoRead

[1]Two Swiss Cantons and many New England towns continue this tradition of direct-democracy, where all key decisions are discussed and put to a vote in large, public meetings.

There is a modern variant called “Deliberative Democracy” or Citizen Juries or mini-publics, where a statistical cross-section of the public is empaneled to provide advice on controversial civic questions. Most famously the Icelandic Constitution of 2011 whose proposed constitution was ignored by the legislature, and the 2016 Irish Citizen Assembly which led to a repeal of the ban on abortion.

Rarely are these juries granted powers beyond an advisory role, but often, by speaking truth to power, they gain influence over the decision makers. Jury members are randomly drawn from volunteers and statistically adjusted to mimic the affected citizenry’s demographic profile. Outside experts provide briefing papers and other advice, with deliberations either streamed to the public in real-time or summarized. Depending on the task, the jury could meet for a weekend or a year.

Criticisms of Citizen Juries include:

• Undue influence of the experts and briefing papers on the outcome.
• The choice of demographic profiles for statistical adjustment could affect the outcome.
• Generally only 1:1000 of the public offers to volunteer, and roughly a third engage in active deliberation. While membership is corrected for obvious categories like sex, geography, wealth, etc these volunteers clearly are not average people, who on average choose not to participate.
• Not all demographic groups are equally skilled or familiar with the tools of debate, and while careful moderation can minimize this effect, the results may skew towards elites.
• Some people view the jury as illegitimate compare to an elected legislature (even though the jury is often larger in size and far more representative by any objective standard. And often spends more time discussing legislation than do legislators).
• Not scalable- a carefully curated process is both time consuming and expensive.
• The recommendations are almost always ignored by their sponsors.

To some extent, these are early days for Direct forms of Democracy in the US. The public has abrogated governing to representatives, and is out of the habit of participating beyond lawn signs and screaming at their TV. It would take a serious, society-wide commitment to make self-governing a habit. Rewards, both financial and sociological, requires investment to help swing cultural norms.

[2] Jefferson, in a June 5, 1824 letter to John Cartwright, appears to second-guess the original choice of an aristocratic Senate by the Constitutional Convention. He writes

“ in the structure of our legislatures we think experience has proved the benefit of subjecting questions to two separate bodies of deliberants; but in constituting these, natural right has been mistaken, some making one of these bodies, and some both, the representatives of property instead of persons; whereas the double deliberation might be as well obtained without any violation of true principle, either by requiring a greater age in one of the bodies, or by electing a proper number of representatives of persons, dividing them by lots into two chambers, and renewing the division at frequent intervals, in order to break up all Cabals.” (my emphasis)

Jefferson may have been inspired by a similar proposal in Condorcet’s Annex to the 1793 French Constitution.

[3] Straw-man concerns over the selection of incompetent citizens dates back to Aristotle. But a moment’s thought should cast doubt on this slander- do you really believe that Attica — for more than two centuries- allowed themselves to be governed by fools? In fact, there were numerous protections at work, many more checks-and-balances than exist in our republican system:

• Only volunteers qualified for sortition. Anyone unsure of their abilities opted out of the process.
• The governing bodies were large (typically 500 or more) and convened in unison, so members benefited from the guidance of other people’s experience and spoke only when confident in their opinion.
• Outgoing magistrates had to account, in writing, for their decisions while in office. Any serving member of government could be sued or investigated for bad behavior (strict scrutiny, e.g. dokimasia, applied). Demagogues could be exiled for a decade (ostracism), while serious fines and other penalties deterred the incompetent. By comparison, Washington is nearly a consequence-free zone…
• If a member spoke in the Assembly about subjects where they possessed no wisdom or competency, they were shouted down.
• A vibrant process to examine, debate and replace bad legislation assured any errors were quickly corrected the following year.
• A few positions with specialized skills, such as Military General or CFO, were filled by appointment or election. There was also a high council with lifetime appointments (the preexisting Areopagus, whose powers waxed and waned). But on the whole, Athenians believed most decisions were best informed by personal experience and intuition, rather than domain expertise.
• A small bureaucracy ensured continuity of operations and institutional memory.

[4] A few additional examples:

• The selective Service draft in times of war
• Nationally, random inspections are an effective means for OSHA, the NRC and IRS to uncover fraud and spur the majority towards honesty.
• By introducing noise, political science research becomes more reliable. Polling results are often skewed by the order in which the questions are introduced. Randomizing their presentation, or varying word choice, eliminates this bias.
• Randomizing the order of candidates or referenda on a ballot eliminates systemic errors and reduces ballot exhaustion.

[5] The first modern example of random jury pools traces back to the “province” of South Carolina in 1682 and was later adopted more broadly in England and the emerging United States.

[6] There are too many virtues of political randomness to illustrate in a short essay. Any list would include:

• Dilutes the power of cliques and factions.
• Makes it harder to game the system via bribery or by controlling the electoral process.
• Encourages a broader range of citizens to participate in government, many with relevant talent but lacking the political instincts to contribute through an election.
• It is well-established that incumbency leads to complacency. Sortition prevents group-think and the premature dismissal of creative ideas. Built-in term limit.
• Athens’ large governing bodies guaranteed sufficient members with subject-matter expertise were always on hand.
• A direct-democracy culture tends to create a more positive view of government amongst its citizens. You can’t run against Washington when you are Washington.
• The process of governing is trusted as neutral (perhaps the same reason there is growing interest in AI agents instead of human legislators).
• Citizens owe their appointment to neither party or patron and can freely speak their mind.
• Even those who do not volunteer for sortition may still attend the general assembly and vote, so their opinion matters at least as much as in a classic representative democracy. Considering many legislators do not reflect the views of their district, but only their party, one can argue a representative democracy stifles half the voices.
• The losing side in a debate has less incentive to pursue un-democratic retribution because they have an equal opportunity to govern after the next lottery. Compared to a representative government, where those in power leverage their position to diminish the chances of their rivals return to office.

It is also worth pointing out that randomness plays an important role in other domains than politics. For example,

• In science and engineering, sometimes adding noise counterintuitively improves measurement accuracy. Take a dial pressure gauge. Dial hands often freeze in the wrong position due to stiction. Tapping on the dial’s window frees the hand, and even though continuous vibration precludes the needle from settling down to a fixed position, its average location is precise.
• In business, ignoring your bosses’ orders can get you fired. Yet, by embracing experimentation, new opportunities appear. A “rogue” McDonald’s franchisee ignored corporate guidelines against early opening and invented the Egg McMuffin, which was so successful, their clandestine novelty was replicated nationwide. Fresh voices lead to fresh opportunities.
• Randomized A/B tests are a marketing best-practice.
• Randomizing promotions (and then sorting out by performance) improves management’s quality and flexibility.
• In surveys where people might lie (e.g. on sexual behavior) humans will respond more accurately if you let them roll a die, and if the die comes up even, randomly answer the question. For odd rolls, answer honestly. A random die offers plausible deniability to the responder, preserves anonymity, and its effect can be subtracted with simple statistics.
• And of course, our very existence as a species depends on random mutations and genetic variability outmaneuvering an impassive Nature “red in tooth and claw”.

While clearly no system is perfect and blending politics with randomness has its downsides, there is much to be learned from the Athenian experiment.

There are also some thoughtful ideas here:

Gastil, John., Wright, Erik Olin. Legislature by Lot: Transformative Designs for Deliberative Governance. United Kingdom: Verso Books, 2019.