A recent book by Nina Roux has the title Tirage au sort et démocratie. Retour sur le cas athénien [Sortition and democracy. Revisiting the case of Athens].
An interview with the author doesn’t turn up that much that is new. Still, it emphasizes some points that may not be new but are worth emphasizing. One point is regarding participation. It is often said that Athens was a radical democracy because of mass participation. Roux points out that this is not the case.
Another important point is about the independence of allotted bodies from elites. In Athens, sortition was not managed by elected institutions or other elite elements. For that reason it could function meaningfully as a democratic force and impose its will on the elites. This stands in contrast with modern proposals and instantiations of allotted bodies (very much including modern juries) which are always concerned keeping those bodies in check.
One of the most stimulating reflections in your work concerns the contributions of the Athenian case to our time. Far from simply observing, you propose a path toward a radical democracy whereby sortition contributes to the people’s control over their leaders. Could you elaborate on this point?
One of the main conclusions of my work was to emphasize that sortition was not the sole source of Athenian democratic radicalism. I observed, in the sources available to me, that there were indeed constraints on citizens who wished to go to the Agora to participate in public debate, and that the city took few steps to alleviate them. To say that Athenian sortition was instituted to guarantee the representation of all social groups in institutions and to create a broad participatory movement is an exaggeration—which does not mean, however, that classical Athens was not a radical democracy. It simply functioned through means of pressure other than the massive, constant, and active participation of all citizens. The most original feature of the Athenian city-state was not sortition, which is found in other institutional contexts: it was the popular courts, composed of thousands of judges chosen by lot, who voted without possibility of appeal on all trials, including those of the city’s politicians. In The Constitution of Athens, Aristotle emphasizes the importance of sortition to protect this institution from corruption and guarantee the effectiveness of popular control over the city’s elites. The use of chance, rotation, and the constant replacement of the set of judges and magistrates prevented this democratic control mechanism from jamming. And it was a mechanism that could choose as its target any politician—even Pericles, deposed in 429 BC for his handling of the first months of the Peloponnesian War. In this respect, sortition goes hand in hand with misthophoria, the remuneration of citizens participating in political institutions. It is also inseparable from the deterrent sanctions imposed by the Athenian courts: fines, death, and exile. It is therefore not insignificant that the allotment machine, the kleroterion, explicitly associated by Aristotle with the fight against corruption, was created in the aftermath of the oligarchic episodes of 411 and 404 BC, which profoundly shook the city.
Filed under: Academia, Athens, Books, Elections, History, Juries, Participation, Press, Sortition |
Adam Cronkright
Ali
Ahmed R. Teleb
avanderven
avergne
Alex Kovner
azakaras
Brett Hennig
campbellwallace
Simon Threlkeld
Nick Coccoma
codyhipskind
Common Lot Sortitionist
Conall Boyle
constitutionalism
davidschecter
Ethan McCutchen
graytmatter
gtridimas
Henry Tuttle
H.J. Hofkirchner
HpR
James Kierstead
John Burnheim
jgarry20151
Jochen Krattenmacher
johnedwardtaylor
johnhuang865
Jonathan Crock
keithsutherland
Jacob Richter
kkoreilly
lukewhito
SimonPek
Nathan El
Nicholas Gruen
Owen Shaffer
Patrick Chalmers
Paul Gölz
peterstone
Pierre-Etienne Vandamme
pronfeld
Emil
rcaze
richwbrown
Roger Knights
roslynfuller
Pierre-Etienne Vandamme
shaharavin
tbouricius
Ted Wachtel
Joshua Laferriere
theodorestathis
towardsautonomyblog
Wayne Liebman
Yoram Gat
Oliver Milne
In her forthcoming book Daniela Cammack agrees that jury courts were a key institution of Athenian democracy (see https://equalitybylot.com/2026/07/30/representation-in-democratic-athens-manins-principles-revisited/
This is one of the reasons that many modern sortition advocates were disappointed by Canevaro’s claim that Hansen’s theory of 4th century legislative courts (nomothetai) was based on a mistranslation. But we should be reassured that even if Athenian lawmaking was the prerogative of the full assembly, modern statistical theory would suggest that employing large samples would have made little difference to the outcome.
See also Ella Street’s paper on just judgment in the Athenian jury courts: https://www.ingentaconnect.com/contentone/imp/hpt/2025/00000046/00000001/art00001
LikeLike
It seems intuitively extremely unwieldy to have a trial with what is effectively hundreds or even more than a thousand judges/jurors.
Are there many accounts of how this was done where we can get a real idea of how it was organised?
Plato’s apology makes it all seem very oratorical.
I realise that attendance was by sortition, but with numbers as high as that, it seems to me that the type of deliberation involved might not be a model of thoughtfulness and might tip over into a susceptibility to demagoguery that seems to have been a feature of the assembly.
LikeLike
Nicholas,
This is something I’ve handled differently in the legislative architecture I’ve been working on. I want the attention and interaction you can get in a small jury, but I don’t want the sampling problem that comes with having eleven people decide for everyone. So instead of enlarging the jury, I replicate it: multiple small independent panels hear the same competing case, deliberate separately if they want to, and vote individually by secret ballot. They are forbidden from communicating outside the group as a standard criminal jury is. The individual ballots are then pooled. This allows for larger numbers without the contagion that seems to occur when groups gets larger.
When a sample size is too small, add panels rather than seats.
LikeLike
Thanks
Yes, that sounds like a good idea, but there are two ways to go about addressing statistical significance. One is to increase the size of bodies, possibly very substantially.
But rather than ramping up sizes, one can ramp up the size of the extent of the (super)majority necessary to carry the day.
Personally, it will be a while before people will regard a small majority of a random selection of citizens as very persuasive even if it’s large and meets statisticians’ ideas of significance.
LikeLike
Nicolas,
People already accept a small number of people representing them. That’s exactly what representative democracy is, right? So, it’s about ‘random’ citizens? If you increase the size of the random sample enough, it becomes virtually indistinguishable from an entire population.
But modern representative democracy has become a religion – not in the spiritual sense, obviously, but it has become faith based. The obvious and overwhelming evidence that it doesn’t work doesn’t seem to convince anyone. The solution seems to be: VOTE HARDER! People talk about Trump followers being a cult, but I see the same willful ignorance in backers of modern representative democracy.
The arguments for elections are far more about mollifying the public than actual reasoning based on facts or historical precedent.
LikeLike
Nick,
The chapter on The People’s Court in Hansen’s Athenian Democracy in the Age of Demosthenes is a good place to start. Hansen claims that the court was the most important organ of state (alongside the assembly), as “the courts had unlimited power to control the Assembly, the Council, the magistrates and the political leaders: political trials were the largest part of the business that came before them” (p. 179). Private lawsuits were judged by panels of 201-401, public prosecutions by panels of 501-2501. Hansen provides no explanation for the numbers involved, other than “a democratic procedure to give every citizen the same chance of selection as a juror, and a bar against attempts at bribery” (p. 197). The Greeks had no mathematical understanding of statistical probability (but that doesn’t rule out an implicit notion of representation).
The system was “accusatorial”, there being no public prosecutor, but both accusation and defence could be divided between several speakers from the same group and, no doubt, the rhetorical style of both parties was oratorical. Most importantly “The jurors had no opportunity for discussion during the casting of the votes: on the contrary, it was thought undesirable for them to exchange even a word or two about the case” (p. 202). According to Ella Street (op cit) “The Athenians did not think that talking with others would serve justice or democratic equality and they prohibited verbal deliberation among jurors in the popular courts.”
I continue to be puzzled as to why deliberative democrats reference classical Athens as an inspiration for citizens’ assemblies etc. It would be much better to use the People’s Court as a model for nomothesia — as, indeed, does Hansen, even if it turned out to be wishful thinking. Henry’s proposal is an interesting compromise, but is complex, potentially expensive and vulnerable to information cascades (as explored by Kai O’Reilly’s paper which he is currently presenting at APSA).
LikeLike
Henry:> People already accept a small number of people representing them. That’s exactly what representative democracy is, right? So, it’s about ‘random’ citizens? If you increase the size of the random sample enough, it becomes virtually indistinguishable from an entire population.
Yes indeed, the principal constraint on sample size (other than cost) being the rational ignorance threshold. But it’s a mistake to conflate statistical sampling and the active representation of interests, where people choose their own spokespersons. This has nothing to do with electoralism as a religion (unless you are suggesting that free will is a religious principle).
LikeLike
Nicholas,
I’d separate the two. A bigger sample is a closer representation of actual public sentiment. A supermajority is about certainty. If you require a supermajority from a small group, you’re mostly just amplifying the randomness of who got picked. Two-thirds of a panel of eleven will reject something with 60% real public support about seven times out of ten. You need enough votes for the threshold to mean anything.
The other problem is that a supermajority usually favors the status quo. That’s defensible for constitutional questions, but applied to everything, little changes, and nobody finds that very persuasive either.
LikeLike
Henry,
I’d contest two of your statements.
I think this very much depends on what it is we are looking to measure. If we take a random sampling of people and simply ask them their preferences, then yes, that seems roughly correct. However, if you are asking them to deliberate and form some degree of consensus, then those thresholds become far more meaningful. If you believe that people, given the right set of conditions, can be convinced by one another, then in those seven out of ten instances, people will need to be convinced for it to pass. Depending on how it’s set up, we could also require that the same thereshold is required for it to be rejected. Given that symmetric design, 60% public support will translate to getting the votes more often than 3/10 of those instances, since there will be less work in convincing a few in the favour to reach the threshold compared to the reverse (assuming people are equally likely to be moved in either direction).
I think this entirely depends, again, on the institutional design. If we are looking at elected officials, I would agree with you whole-heartedly, as they have external motivations that prevent them from making concessions. In fact, in many instnaces it is in their best interests not to work alongside other parties, or to do so only when they can get something in return.
A jury system, that we tend to agree upon, lacks those external motivations. While some people may benefit from a piece of legislation, while others may be harmed by it, both of them are intrinsically motivated by the legislation itself, while others may not have a strong preference at all. This frees them to balance their own benefits and drawbacks to the benefits and drawbacks of others, and potentially make conecessions if they feel it worthy. I think they are far more likley to do so than elected officials, who, upon making a concession of this nature, risk losing an election.
LikeLike
Jordan,
On the second point, I agree that elected officials have a whole host of external motivations. My entire belief system is based on every institution having a multitude of motivations, and the longer they’re in place, the more they functionally deform the body. Legislative gridlock really is mostly about elections — credit-claiming, primaries, not wanting to hand the other side a win. Jurors have none of that, and I agree they’d be more willing to move than a legislator would.
But I don’t think status quo bias is about obstruction. Nobody needs a motivation to leave things alone. Given a choice between what they know works well enough and something new, most people stay put unless there’s strong evidence or strong feeling pushing them the other way. That isn’t a legislator protecting his seat, it’s just how people decide. Requiring two-thirds to change something asks a body with no particular stake in acting to clear a high bar for action, and I’d expect that to favor what already exists.
I had AI run the numbers. At eleven jurors with a two-thirds rule, a bill with 50% real public support passes 11% of the time, and one with 80% support fails 16% of the time. The rule is supposed to mean two-thirds of the public would agree. At eleven votes it can’t tell 50 from 80.
Which brings me to deliberation. My panels do deliberate. What I don’t endorse is consensus. Consensus is usually someone giving in rather than someone being convinced, and the one who gives in is generally the person with the weaker opinion rather than the weaker argument. Deliberation is useful for surfacing facts and clearing up confusion. I just don’t want the vote turning on who outlasted whom in the room.
On requiring two-thirds to reject as well — that rests on people being equally movable in either direction, and I don’t think they are, for the reason above. The bigger issue is the middle. At eleven jurors and 60% real support, it passes 30% of the time, rejects 3%, and two-thirds of the time reaches neither. So it needs deliberation to move people in most cases, not a few. And if the room can’t finish until it hits a threshold, whoever holds the weaker opinion has to fold. That’s the giving-in I just described, except now the design requires it.
Where we agree: thresholds should scale with stakes. My objection is only where the threshold sits. At 660 pooled ballots a two-thirds rule passes what has two-thirds support and almost nothing else. At eleven it’s mostly measuring the draw.
LikeLike
Henry,
Do you think that consensus always works that way? Do you think that sometimes people can be genuinely convinced?
Can it be sometimes that a person may not agree with the consensus, but is at least convinced that there is enough merit to it that they could live with it? Would this state of mind have value to it?
In a group of randomly selected citizens, a good deal of them will have very weakly held opinions, and can be convinced in either direction depending on the strength of the arguments.
I think there is actually a great deal of evidence that given the right set is conditions, people can form a consensus without people being steamrolled.
And again, the numbers you are providing from odds are assuming that people will simply vote based on the position they begin at. I’m unconvinced, at this moment, that that is a reasonable assumption to make.
LikeLike
>>Do you think that consensus always works that way? Do you think that sometimes people can be genuinely convinced?
I do think that sometimes people can be convinced. But I SUSPECT that “I’m not convinced, but I don’t strongly disagree and I just want to go home” happens MORE OFTEN than being convinced. “Steamrolling” was too strong a word for that. But I haven’t found any empirical evidence that establishes which happens more often.
Thing is: in general, I’ve found that many people who claim to be convinced (all anecdotal) have to be convinced of the same thing at a later date. They’ll say “yeah, I see your point!” And then a few months later, make the same exact argument that they claimed to have understood the error of. That doesn’t prove they weren’t convinced initially, but it makes me question how much that apparent agreement tells us.
And yes, “I could live with it” can have value. I don’t think someone has to consider a proposal their first choice to sincerely support it. In fact, “I could live with it” is a choice category in my jury panel along with “I strongly endorse this” and “No”.
I DO believe in deliberation. I think it ABSOLUTELY clarifies thinking in many ways.
But some of the strongest evidence for beneficial opinion change comes from deliberation that doesn’t require consensus. So my question is what REQUIRING consensus adds. People can be convinced, or decide something is an acceptable compromise, without making agreement a condition of finishing. My concern is the extra pressure that condition introduces.
Also, it obviously depends on the topic. I think you are going to find it much easier to find consensus on whether Joan in accounting should dump her boyfriend than whether Jesus is the savior—or even more polarizing, whether elections are a good idea.
LikeLike