A The struggle between oligarchs and democrats, an overview

The Athenian democracy collapsed twice at the close of the fifth century. In both cases, it was replaced by an oligarchy that promptly ignored legal rules on a wide scale. Strikingly, both collapses immediately followed an exogenous military shock.

The first happened right after Athens’s notoriously ill-advised invasion of Sicily.² After the military adventure collapsed, Alcibiades, from exile (thanks to the affair of the Herms/Mysteries), attempted to provoke a coup. Conspiring with Alcibiades, Peisander convinced the assembly to accept unspecified restrictions on the democratic franchise, negotiate with Alcibiades for his potential recall, and appoint a commission (the syngrapheis) to investigate the state of the city. This, on his argument, would convince the Persian king to lend financial support to the continued prosecution of the war against Sparta, the Sicilian adventure having put the city into serious financial straits. On Peisander’s instructions, oligarchic clubs (hetaireiai) within the city began a campaign of terror and intimidation, carrying out several assassinations, including at least one democrat prominent enough for Thucydides to describe him as “the chief leader of the people.”³

According to Thucydides, this campaign of terror worked: fear of hidden conspirators inhibited democrats from speaking up at the assembly or trusting one another enough to carry out collective action.⁴ The syngrapheis proposed the abolition of the graphe paranomon and the transfer of authority into the hands of 5,000 citizens. Meanwhile, Peisander claimed that the Persians demanded a still smaller oligarchy (actually, he knew that Persian support was not forthcoming), then proposed the Four Hundred. The assembly was intimidated into compliance, and the Four Hundred took office, drove out the democratic council by force, and assumed power.

Perhaps predictably, the Four Hundred promptly began to ignore the rule of law. According to Thucydides, they “ruled the city by force; putting to death some men though not many, whom they thought it convenient to remove, and imprisoning and banishing others.”⁵

The Four Hundred didn’t last long. They had a big problem: the Athenian navy was at Samos, and it was “dominated by the lower classes.”⁶ In order to shore up their position, they repeatedly tried negotiating with Sparta, and also reendorsed their earlier promise to extend citizenship to 5,000 citizens. The promise was not enough to satisfy the mass opposition, and the Spartans, rightly mistrusting the stability of the regime, preferred to take advantage of the chaos and launch an invasion rather than make a deal with the oligarchs.⁷ With a Spartan fleet at the door, the oligarchy promptly collapsed, being first replaced by the promised rule of the Five Thousand, then, shortly thereafter, a restored democracy. Under the latter, a number of oligarchs were tried and convicted of treason and subverting the democracy.⁸

The Thirty, despite its extensive overlapping personnel with the Four Hundred, originated and operated very differently.⁹ It was imposed by Sparta after the final Athenian defeat in the Peloponnesian War, and was, on Xenophon’s account, initially welcomed.¹⁰ However, the Thirty quickly went bad. They started by surrounding themselves with whip-bearing guards (always a bad sign). They reallocated the function of the people’s courts to a puppet council. They carved out 3,000 elites to remain full-fledged citizens, and enacted a law permitting the Thirty to kill any of the rest at will. They disarmed the non-3,000 and forbade them from remaining in the city limits. They stole a lot of property.¹¹ The Thirty are generally credited with about 1,500 murders.¹²

Thrasybulus, an exiled Athenian general, led a revolution. The Thirty called for Spartan aid, but the Spartan king commanding the relief troops grew tired of the trouble and imposed a peace on the warring parties.

The terms of the peace, in summary, were as follows: the democracy was restored, but all of the oligarchic party except the actual Thirty (and a couple of other small, irrelevant groups) were to be given amnesty for all their crimes except personal murders. The Thirty themselves were to be subjected to euthynai, with a small thumb on the scale in their favor (the jurors were limited to property owners), and would be rehabilitated after accepting whatever punishment the court imposed. (At least one member of the Thirty passed this examination, and returned to citizen life.¹³) Unsold expropriated property was to be returned to its rightful owners. And those oligarchs who wished to do so were to be allowed to exile themselves to Eleusis instead. The amnesty was, on the whole, obeyed.¹⁴

B The puzzle of the amnesty

The Four Hundred, according to Thucydides, extradjudicially killed “not many” people.¹⁵ Taylor has argued that the role of violence and terror in their coming to power has also been exaggerated.¹⁶ Compared to the Thirty, the rule of the Four Hundred seems to have been characterized by a remarkable restraint in the murder, robbery, imprisoning, and exiling departments.¹⁷ Yet, the Thirty received the benefits of an amnesty, while the Four Hundred were prosecuted. The amnesty was imposed by Spartan swords, but not enforced by them. Why did the democrats, dominating the assembly and courts, not promptly repudiate the amnesty and then execute the oligarchs (with or without trial)?

Two strategic hypotheses come immediately to mind, but neither is convincing. First, the democrats may have feared the return of Sparta to protect their oligarchic political allies. However, Athens joined the Corinthian war against Sparta less than 10 years after the Thirty were deposed.¹⁸ Executing a few oligarchs doubtless would have annoyed the Spartans less than going to war against them did.¹⁹

Second, the establishment of an oligarchic state-in-exile in Eleusis may have been meant to provide the oligarchic party with enough resources to credibly threaten retaliation should the amnesty be violated. However, oligarchic Eleusis did not last very long: it was swiftly reconquered and reintegrated into Athens proper.²⁰

David Teegarden argues that the actual oath taken to uphold the amnesty made it possible for the community to avoid private violence against former oligarchic collaborators, because it generated common knowledge of, in his words, citizens’ “(at least apparent[ ]) credible commitment” not to retaliate against collaborators.²¹ By doing so, it gave individual Athenians who might otherwise want to retaliate some reason to think that their fellow citizens would not support them. Since retaliating against collaborators was individually risky, they would not be willing to do so if they believed they would be unable to count on the support of their fellows.

However, Teegarden’s argument makes the oath do too much work. As a general principle (albeit with a number of exceptions), mere costless words cannot establish a credible commitment; instead, they often are nothing more than cheap talk that does not change the underlying strategic dynamics of a situation.²² In the Athenian context, the cheap talk interpretation of the oath seems most plausible. The amnesty and oath were imposed at sword point by the Spartan army. Under such circumstances, vindictive democrats would have had little reason to believe that the oath represented their fellow citizens’ true intentions or preferences.

The traditional explanation for the success of the amnesty has been nonstrategic. Ostwald summarizes classical opinion as varying between “the patriotism of the Athenians as a whole” and “the forbearance and decency” of the democrats.²³

The “forbearance and decency” argument ignores the fact that Athenian democrats did retaliate (in the courts) against the less grievous crimes of the Four Hundred only a few years beforehand. It is unlikely that the Athenian democrats experienced a collective cultural or ethical change between 411 and 403. Moreover, it is inconsistent with the fact that the democrats even retaliated against collaborators with the Thirty, just not in ways forbidden by the amnesty. The cavalry, for example, was a military role occupied in Athens by relatively wealthy citizens, and whose members largely supported the Thirty. After the Thirty, the democrats lashed out at the cavalry twice: first, by cutting their pay in order to raise the pay of the lower-class archers, and second, by deliberately sending 300 of them off to die in a foreign war.²⁴

With the “forbearance and decency” argument ruled out, the most plausible explanation for the success of the amnesty in the existing literature is Lanni’s, which comprises four elements.²⁵ First, she argues that there was a postwar process of whitewashing in the courts that focused blame for the tyranny on the Thirty themselves rather than on their many collaborators. On her account, litigants adopted this strategy on an individual basis, presumably because it would be most palatable to the jurors, many of whom would have been collaborators themselves. Despite that history of collaboration, Lanni points out that the forensic speeches often addressed the jurors as if each member had been a part of the resistance. The ultimate effect of this strategy was to construct a false “collective memory” in which most ordinary citizens were innocent of crimes under the Thirty.

Second, Lanni notes that litigants often used the amnesty as an example to illustrate the mild and virtuous democratic character of the Athenian people. Consequently, she argues, the Athenians came to collectively identify as the sort of people who offer amnesty to their enemies, and to become motivated to continue doing so.

Third, Lanni points out that the amnesty contained a “safety valve” for individual cases: because crimes under the Thirty could be raised as character evidence in unrelated cases and in dokimasiai for incoming magistrates, limited-scope accountability was allowed. This satisfied some of the desire for revenge before it could spill over into a movement for broad-brush retaliation.

Finally, Lanni suggests that Athens’s participatory political institutions may have, by forcing former oligarchs and democrats to work together, given them reason to repair their relationship after the oligarchy.

Lanni’s account is partially convincing. But the material given thus far allows us to supplement it with an additional explanatory factor. The development of the law through and after the time of the oligarchic revolutions is consistent with the increasing recognition of the importance of law for the stability of the democratic state – the strength topos. The law reforms of the postconflict period suggest that the consciousness shown in the evidence for the strength topos was growing at that time. An effort had already begun to collect and codify the laws at the time of the Four Hundred, and after the Thirty, as noted, the democrats further strengthened their legal system by creating the quasi-constitutional difference between laws and decrees, requiring all acts of the assembly to be scrutinized against the existing law code and similar reforms.

Moreover, as Cohen cogently argues, the Thirty came to stand for grievous violations of the law in Athenian political culture. Democratic politicians, by contrast, laid claim to institutions of the rule of law in order to “bind the community together in opposition to its oligarchic opponents who sought to undermine its institutions to create stasis [factional conflict].”²⁶ The institutions of the democracy, including those legal institutions that the Thirty disregarded, became, on Cohen’s account, identified with the democracy in part because the democracy identified itself in opposition to the Thirty, and the Thirty saliently disregarded the laws.

I submit, then, that the democratic obedience to the amnesty reflected a developing respect for the law among the Athenian people. The Athenians came to identify the law with the democracy and the equality that it represented (collectively, as isonomia), at the same time as they came to the belief that careful compliance with the laws was necessary to their political strength and stability.²⁷

And the Athenians were correct to see it that way. The law could preserve the strength of each individual Athenian in the face of elite power by coordinating resistance to elite hubris as well as to outright threats to undermine democratic institutions. Athenians essentially were in a game-theoretic coordination equilibrium in which each knew that his fellow citizens would resist any illegal acts; this gave nonelite citizens the ability to rely on the law, embodied by their fellow citizens on the jury, to defend them from the elites. However, for the law to serve this function, each citizen must have believed that his fellow citizens would enforce the law. Since disregarding the amnesty would indicate jurors’ willingness to throw aside the law in favor of political expediency, it would have vitiated this coordination function: no longer could citizens trust in the strength of the law to defend themselves from oligarchic hubris. And this is why moderate democrats were correct to see the jury, rather than the assembly, as the chief institution of democracy.

To anticipate an objection: Athenians had to rely on law to serve this function, rather than simply sharing a commitment to resist oligarchic acts, legal or illegal, because of the potential for uncertainty as to whether any given act posed oligarchic dangers. Frequently in the forensic speeches we see elites accusing one another of oligarchic sentiments and identifying themselves with the masses; this suggests that both sides of a legal dispute could often be plausibly characterized as oligarchic. But if someone were caught breaking the law, this could serve as an objective sign that the malefactor held an inadequate regard for the democracy. Moreover, a jury verdict could serve as a consensus signal of guilt on which citizens could rely to coordinate their opposition to an overweening potential oligarch. If a majority of a large and socially representative jury working in the glare of publicity was willing to condemn someone, each individual in the city could infer that the community at large would be similarly willing.²⁸ Thus, the law allowed citizens to infer oligarchic threats from a verdict, and provided common knowledge that each democratically inclined citizen would be willing to resist that threat: it was the vital keystone for civic trust.

In short, the amnesty worked because the Athenians developed a public commitment to the rule of law, in their official capacities as magistrates, jurors, prosecutors, and assemblymen.²⁹ If this is right, then Lanni errs in asserting that “the absence of the rule of law is a feature of the system [for promoting Athenian reconciliation after the Thirty] rather than a bug.”³⁰ On the contrary, the Athenian rule of law, rooted as it was in citizens’ recognition that the control of power by law was a precondition of an equal state, was a vital part of the success of the amnesty. Moreover, the commitment to law was useful not only to the masses, but also to the elites. Even those with oligarchic sympathies, who had committed crimes under the Thirty, would have reason to uphold the law – even to the extent of prosecuting their political allies for future oligarchic crimes – because if the law collapsed, the amnesty would cease to hold, and there would be nothing protecting them from mass vengeance. This is quite a trick: the law could reconcile the elites to the legal system by protecting them from punishment for their past misconduct, and at the very same time, to the extent the masses actually extended that legal protection to elites, it strengthened their own power to prevent future elite misconduct.

The democrats exactly recognized Sir Thomas More’s worry, in A Man for All Seasons, that should they chop down all the laws to get at the oligarchic devil, there would be nowhere for them to hide when the oligarchs turned around and went after them.³¹ If any doubt remains on this point, consider that none other than Thucydides himself originated that particular insight:

For this reason, the democrats were concerned to preserve the Athenian rule of law, and were willing even to sacrifice their revenge against the Thirty Tyrants to do it.

C Did the Athenians learn from experience?

There is some reason to believe that the strength topos and its invocation in defense of the amnesty reflected the Athenian democrats’ experience in the periods leading up to the two oligarchic coups. The material given thus far suggests the hypothesis that previous failures of the rule of law may have contributed to the initial success of both coups.

It’s striking that the two major collapses of the democracy happened not only after major military defeats, but also after major lapses of the rule of law. The affair of the Herms/Mysteries, in which many citizens were executed or fled into exile on very scanty (and later discredited) evidence, happened at the beginning of the Sicilian expedition. That expedition precipitated the coup of the Four Hundred, and Alcibiades was one of the targets of the Herms/Mysteries witch hunt. Similarly, the coup of the Thirty was preceded by the trial of the Arginusae generals. These correlations may result from causation: if the affair of the Herms/Mysteries and the trial of the generals sufficiently undermined citizens’ confidence in their fellows’ willingness to follow the law under exigent circumstances, that may have contributed to their failure to do so at the time of the coups.³³

This hypothesis draws some support from Thucydides’ description of how the terror tactics leading up to the coup of the Four Hundred worked:

That is, on Thucydides’ account, the rise of the Four Hundred was attributable in large part to the decline of civic trust among the Athenians, and that decline in civic trust made them unable to use the legal system to put a stop to oligarchic threats. This fits nicely into the causal hypothesis I’ve suggested: perhaps the Athenians ceased to trust their legal system (at least in part) because they recognized that their fellow citizens couldn’t be relied upon to enforce the law in times of crisis, and that recognition was in turn based (at least in part) on their shameful behavior four years before in the affair of the Herms/Mysteries. It also fits the origin of the Four Hundred – in a council ostensibly called at first to restore the traditional laws.³⁵

Moreover, Thucydides seems to be suggesting a broader decline in civic trust – not just that citizens failed to trust the legal system, but that they failed to trust one another in general. Contemporary empirical evidence exists to support the hypothesis that such a broader decline in civic trust, or “social capital,” could be due to the flaws of the legal system: a recent study has suggested that regions that had the advantages of impartial and reliable legal institutions, in the form of the Napoleonic Code through the nineteenth century, show greater social capital even today.³⁶ Thucydides appears to be describing the contrapositive of that effect: with the failure of the legal system, the Athenian democrats lost the social capital that could have helped them collectively resist the Four Hundred.³⁷

Contemporaneous sources confirm my supposition with respect to the rise of the Thirty Tyrants. On Xenophon’s account, one of the first acts of the Thirty was to summarily execute those who were alleged to be “sycophants” – the equivalent of modern professional frivolous litigators, the Athenian “ambulance chaser.”³⁸ This first bloodletting met with universal approval.³⁹ Regardless of whether sycophants were actually a problem, Xenophon clearly expected his readers to believe that the Athenian public thought the legal system was being routinely abused.

At least one scholar has further suggested that the alleged sycophantic problem at the time of the Thirty arose out of attempts to retaliate against those who were attached to the Four Hundred.⁴⁰ According to Jordović, sycophants operated by bringing litigation against innocent aristocrats, to target them in a widespread “settling of scores” with the Four Hundred (presumably by falsely accusing those innocents of being part of the conspiracy). Jordović has some support in Lysias, one of whose clients (for Lysias’s speeches were written for the use of others) suggests that after the fall of the Four Hundred, demagogues “persuaded [the people] to condemn some people to death without trial, to confiscate unjustly the property of many more, and to expel others and deprive them of citizen rights,” and goes on to say that this “reduced the city to civil strife and very great disaster.”⁴¹ On Lysias’s client’s account, “oligarchy has twice been established because of those who were sycophants under the democracy.”⁴² If Jordović and Lysias’s client are right, the zeal for retaliation after the first oligarchy helped bring about the second oligarchy, by undermining citizens’ confidence in the legal system and winning public support for the first round of tyrannical executions.⁴³

No surprise, then, that the strength topos began to get a grip in the public legal and political culture of Athens after the fall of the Thirty. Perhaps the success of the amnesty came about because the democrats learned from their prior mistakes.⁴⁴ Put differently, the Athenian citizens developed a preference for law over short-term political advantage.

D The problems of commitment: disagreement and temptation

But we must pause here and add something to the strength topos explanation. The mere recognition among those (elites) who were producing the oratorical corpus, history, poetry, and philosophy that the law was important to the strength of the democracy is not enough to fully explain the amnesty’s success. For the citizens must also have trusted one another to share that recognition, and this must have been particularly difficult in the face of their dubious recent history of throwing aside the law in panicked lashing out at suspected oligarchs.

Particularly worryingly, there must have been a significant temptation to ignore the amnesty in the short run. And this temptation would be self-reinforcing: if the amnesty wasn’t reliable because a citizen’s fellows were subject to temptation, then the given citizen ought to give in to temptation himself and get rid of as many oligarchs as possible. Suppose an oligarch was facing trial for the crimes he committed. A citizen on the jury would have had good reason to worry: would the law be functioning tomorrow? If not, it would be better to eliminate the oligarch on the spot, in anticipation of future class conflict. Moreover, a citizen might worry that his fellow jurors were inadequately representative of the policy preferences of the public: suppose a majority happened to be political opponents of the amnesty?⁴⁵ Furthermore, who’s to say that the jurors didn’t see a given particular oligarch on trial at the given moment as sufficiently dangerous to justify ignoring the amnesty just once? How did the Athenians manage to, in the jargon of political science, credibly commit to the amnesty: to enforce it in support of their long-run interests even in the face of a short-run desire by aggrieved democrats to the contrary?⁴⁶

Moreover, in order to successfully fight off future oligarchic threats, the democrats of Athens must have learned to trust one another. The elite, like all elites, were more powerful than members of the mass on a one-to-one basis; they had the capacity to do things like bribe or intimidate assemblies and juries. They also had greater capacity to coordinate their own actions to magnify their individual power, due to their smaller population and preexisting organizational capacity in the form of the hetaireiai.⁴⁷

In order to counteract these advantages, the masses would have had to make full use of the only advantage they had: numbers. They would have to have been able to trust one another for support to resist future elite coups, even though elites would have had the power to do them harm on a one-to-one basis. And this need for support is reciprocal and extends across the political community. A depends on B’s support, but B, in order to be able to support A, depends on A’s reciprocal support, and both also depend on the support of citizens C and D: each individual democratic, nonelite citizen, in order to be willing to stand up to a member of the oligarchic elite, must be able to rely on the support of each other individual democratic citizen. Otherwise, doing so is too dangerous. And it gets worse. Thucydides tells us that they were unable to so trust one another when the Four Hundred took over, suggesting that they must have had to build this trust out of nothing after the Thirty.

The democracy had a coordination problem: each nonelite citizen rationally should have done what each other nonelite citizen was doing. He would want to punish future elite transgressions if and only if each other citizen could be counted upon to do so – then he would get his most preferred outcome, a functioning democracy. However, he would want to refrain from attempting to punish elite transgressions if other citizens could not be counted upon to do so, because so refraining would allow him to avoid his least preferred outcome, being crushed by an overwhelmingly powerful member of the elite, although it would force him to accept his second least preferred outcome, living under an oligarchy.⁴⁸

Political scientists have shown that law, understood as a common-knowledge mapping of conduct to evaluations (“legal” or “illegal”), with some mechanism for producing authoritative decisions about those mappings, can facilitate coordinated sanctioning systems.⁴⁹ However, in order to achieve coordinated punishment against some transgressor, the democratic masses must have had some settled way of determining when a transgression had occurred. And they must have been able to trust that one another would apply it faithfully. The upshot is that the Athenian democracy after the Thirty Tyrants critically depended on each citizen’s ability to predict each other citizen’s behavior; the law was the instrument for this prediction just to the extent that citizens knew one another could be depended on to uphold it.⁵⁰ But from where came this knowledge?

I contend that the mass jury served these functions. First, it provided the necessary authoritative resolution of disagreements about whether a given course of conduct was sanctionable under law, and thus made it possible for the democracy to coordinate itself by law in the first place. Second, it is this jury that would have heard charges against oligarchs, both amnesty-violating charges for crimes committed under the Thirty Tyrants, as well as charges for new oligarchic crimes committed after the Thirty; it is also the jury that would have heard charges against self-help amnesty violators, had any (other than the one we know of) happened. And – this is the key point – this gave the jurors reason to support the amnesty, because by doing so they could signal their willingness to one another to stand up for the law, and hence solve their coordination problem, and protect their democracy.

Because it was a mass institution, filled with hundreds of randomly selected citizens, and because trials were held in public, the jury could serve as an excellent informational proxy for the extent to which the citizen body was willing to enforce the laws, notwithstanding differences in policy preferences over the amnesty. A jury that ruled for a clearly legally correct outcome, especially when that outcome was contrary to the short-run self-interests or preferences of many of its members, and where that jury was drawn from a large and fairly representative sample of the population, demonstrated a widespread commitment to enforcing the laws and, consequently, the extent to which the laws were likely to be enforced in the future. Finally, because individual votes were secret, each individual citizen was not in danger of retaliation from voting to enforce the law, even if the vote was against the interests of powerful members of society. In short, the jury could build a record of lawful behavior without presupposing it: by safely allowing citizens to demonstrate their willingness to support one another and the law with their votes in the jury room, it allowed them to trust one another enough to be willing to take the risk to do so elsewhere, and hence allowed them to credibly threaten to resist future oligarchic coups in the streets as well as in the courts. The mass jury served the dual role of resolving legal disputes and demonstrating that the populace was committed to following the law.

This is a point similar to one that Ober has made about alignment in Athens in general.⁵¹ As Ober suggested, because the jury was secret and nondeliberative, it could aggregate independent knowledge of individuals; in line with his account I’d emphasize that the knowledge it aggregated included not just facts about the world but also facts about individual preferences, particularly democratic legal preferences.

Moreover, the jury effectively eliminated the coordination problem in amnesty cases: a citizen voting (unlike, say, a citizen taking up arms) need not do what each other citizen does. If he votes, in violation of the amnesty, to convict an oligarch for prerestoration crimes, a majority of his fellow citizens can either also vote to convict, in which case at least the oligarch doesn’t go free, or they can vote to acquit, in which case the amnesty is upheld. The same is true if he votes to acquit. In either case, the worst-case option, in which the rule of law fails and the citizen gets punished for trying, and failing, to kill an oligarch, is eliminated by operation of the secret ballot and aggregative mechanism.

This was at least sometimes a costly signal (unlike Teegarden’s oath-taking), when democrats had to swallow what would otherwise very likely be their preferences for executing any given troublesome oligarch in order to uphold the law. Accordingly, their willingness to do so could serve as a credible signal of their conviction in the strength topos, and thus their willingness to uphold the law in general. For that reason, the amnesty was self-reinforcing: by obliging democrats to act against their short-run preferences, it enabled them to signal commitment to the law; this in turn established a record of consistent conduct upon which each citizen could rely in predicting the behavior of his fellow citizens; this in turn made it less risky for each democrat to use the legal system to resist any oligarchic threats that might arise. And this, in turn, suppressed those threats: in the jargon of game theory, attempting future oligarchic coups was off the equilibrium path, because the democrats could credibly threaten to collectively respond with overwhelming force against any such attempts.

On this analysis, the power of the demos and the power of law are fully reconciled through the strength topos. Prior scholars have sometimes thought that some of the references to what I call the strength topos in the oratorical corpus reflect a tension between the two. For example, Allen cites the following assertion from Aeschines as evidence against the Athenian rule of law: “The private man rules as king in a democratic city by virtue of the law and his vote.” Allen suggests that this represents “some ambivalence over whether the laws or the jurors can ultimately be said to rule the city [because] [t]he citizen’s vote must be given as much weight as the law.”⁵²

But there can be no choice between jurors and law, for the power of each was necessarily interdependent. Aeschines understood that the law ruled through citizens’ votes, and citizens ruled through the existence of a well-functioning legal system. In fact, that passage is a near-perfect statement of the strength topos and the role of the law in constraining the powerful as well as – as I shall further discuss in Chapter 8 – the necessity of the rule of law to democracy. Aeschines goes on to argue that jurors make a grievous error in casting their votes for the interests of powerful politicians rather than the law: first, because the politicians cannot reward them (their votes being secret – and this is another way in which the jury was functional for encouraging people to signal legal preferences), but more importantly, because doing so encourages the rhetors in their hubris and threatens a recurrence of the Thirty. That is, Aeschines is telling the jurors to preserve their political power through fidelity to law. There is no ambivalence here: the law and the jurors must continue to rule the city together. And how could it be otherwise? As I discuss in the next chapter, no legal system can “rule” alone: in every society, there is a group of people who could overthrow it if they cease to be aligned to it; they are the rulers behind the rule of law; to observe that state power is reliably constrained is just to observe that there are strong institutional and political barriers to their alignment except through and under the terms of law. The Athenians were merely unusually perceptive in having seen this, and unusually honest in having said so.

E Athens as a case of transitional justice

“Transitional justice,” in the sense used by contemporary human rights scholars, covers those legal and political mechanisms that promote community reconciliation in new or restored democracies after conflict, including accountability or amnesty for those who were responsible for atrocities in the previous regime. The Athenian reconstructions after the oligarchies of the Four Hundred and the Thirty were perhaps the first instances of transitional justice in recorded history, as many scholars have recognized.

Some have suggested that transitional justice arrangements should aim to publicly and collectively reaffirm the values of the society in question (e.g., to display and demonstrate disapproval of the crimes of the previous regime), in order, inter alia, to rebuild “civic trust” that those norms will be enforced in the future.⁵³ I have suggested that the Athenian jury was functional for this purpose, insofar as it allowed citizens to send signals of their willingness to enforce the law. However, ordinary citizen-initiated litigation was not the only Athenian legal practice to serve this function.

Athens also had something analogous to today’s truth and reconciliation commissions, which seem to be a central feature of what we today label “transitional justice.” The Thirty were required to subject themselves to formal euthunai, examination of former officeholders, in order to return to Athenian citizenship. And if collaborators wanted to take office under the restored democracy, they, like all candidates for office, were subject to dokimasia, formal examination of a prospective official’s character, at which their crimes could be considered. Lanni argues that those institutions operated as a “safety valve for local resentments,” allowing some measure of revenge to be exacted.⁵⁴ The analysis in this chapter suggests that her account is correct but incomplete: these measures also gave citizens an opportunity to choose the lawful method rather than some other method to seek accountability for past crimes, and hence to affirmatively and publicly show their commitment to the accountability mechanisms provided by law.

This similarity is striking. Understanding the function of the Athenian version of the truth and reconciliation commission in the context of the jury and the function of rebuilding trust and coordination may allow us to further understand the function of such institutions in the contemporary world. The functional isomorphism between Athenian institutions and contemporary ones suggests that such an approach may hold substantial promise.⁵⁵

A The model

Start with a lawless political community. Assume, for simplicity, a one-dimensional policy space, which can be approximated as a scale from elite-preferring to mass-preferring policy. Such a policy space can be modeled as a division of a fixed surplus of goods, G (which can include things like political rights as well as economic resources), over a recipient class composed of N people, where the policy space indexes N (e.g., on one end is a pure extractive despotism [N = 1] or a regime like the Thirty Tyrants [N = 3,000], on the other is egalitarian utopia [N = population (P)], and in between are, e.g., aristocracies, societies like Athens that provide equal rights to citizen males but fewer to others, etc.).

The game is structured as follows. Each round, the ruler or ruling group (which models officialdom in general, acting in concert) chooses N, and then each citizen chooses whether or not to revolt. (Revolting can stand in for a wide variety of ways to resist and sanction rulers with no loss to the model.) If citizens do not revolt, the policy is implemented, and each citizen in N (including the ruler/ruling group) receives G/N, while each other citizen receives 0. If citizens do revolt, each citizen who revolts will pay a cost of C, and the revolt will succeed with probability KR, where K is a constant and R is the number of citizens revolting. Each citizen receives a payoff of G/P and the citizens impose an additional cost of F on the ruler if they succeed. If they fail, the policy is just implemented.⁶¹ Note that in this sketch, C models both direct and retaliation costs, and the difference between G/N and G/P represents preference costs for those citizens in the preferred group.

Absent some kind of commitment mechanism of the sort to be outlined in a moment, citizens within N have no reason to revolt. Consequently, the upper limit of R is P – N. Under these assumptions, a citizen within the excluded group will prefer revolting to acquiescing in a ruler’s policy when KRG/P – C > 0. Rearranging terms, this will be when R > PC/KG.

Suppose it is common knowledge that each excluded citizen will revolt when P – N > PC/KG (in words, when enough citizens have been excluded to make it worthwhile to revolt). Call this the common knowledge condition. In such a case, R = P – N.

Under the common knowledge condition, the ruler has an incentive to set N at the smallest number (maximizing his own payoff) that makes excluded citizens indifferent between acquiescing and revolting (perhaps plus one), P – N = PC/KG, or, rearranging terms, N = P – PC/KG.⁶²

The problem is that the common knowledge condition is likely to be false in most societies, because revolting is a public good. To see this, suppose the ruler sets N = P – PC/KG – 1,000. Rearranging, P – N = PC/KG + 1,000. Now suppose P – N citizens revolt. The expected value of that revolt is positive relative to the situation in which no citizens revolt. But suppose one citizen declines to revolt. That citizen saves the cost C, and reduces the probability of success slightly: his payoff (the shirking payoff) is K(R – 1)G/P. The shirking payoff will be higher than the payoff for participating in the revolt when K(R – 1)G/P > KRG/P – C, which is true, after simplifying, whenever C > KG/P. Under those circumstances, our citizen prefers to shirk and free-ride on the provision of revolt by other excluded citizens. Each other excluded citizen, of course, can make a similar calculation. This reveals a classic public goods problem among the pool of excluded citizens. By induction, we would expect a maximum of PC/KG citizens to revolt.

This, in turn, poses serious problems related to selecting which group of PC/KG citizens, among the many possible subsets of the full group of P – N citizens available, will pay the cost of revolting. Consider the risk of strategic behavior: if exactly P – N citizens are poised to revolt, a citizen within that group has an incentive to threaten to shirk in order to force some other shirker to act. In sequential play, this problem goes away: it’s trivial to predict that every citizen but the last PC/KG citizens shirks, and then the rest are left holding the bag. But in simultaneous play, the solution is much less clear; arranging the group of citizens who actually revolt is a difficult coordination problem.

Problems become much worse if we relax the implicit perfect information assumption of the discussion thus far. Suppose C is not uniform, but a private constant for each citizen representing, realistically, facts about that citizen such as his or her risk aversion, idiosyncratic preference costs, and the like. A citizen’s C will determine the number of other participants for which it is worthwhile for her to revolt; without knowing C for each other citizen, it becomes impossible for any given citizen to predict the number of people who will revolt at any given N. Again, in sequential play, the problem might be resolved: Kuran’s revolutionary cascades are one way, for appropriate distributions of C.⁶³ But matters are a mess in simultaneous decision contexts (that is, where citizens cannot observe one another’s choices before themselves choosing, either because they actually act simultaneously, or because they simply lack the relevant information), or where C is not distributed such that the conditions are right for a revolutionary cascade (e.g., if there are no citizens with sufficiently low C that they might start a cascade).

To solve these problems, let us zoom out to multidimensional policy space. Suppose there is a common-knowledge set of criteria (aka “law”) for acceptable policy along a number of dimensions, where each dimension represents an ordering of the population according to some different criterion (gender, race, socioeconomic status, land ownership, regional origin, religion, etc., as locally appropriate) and has some minimum lawful N along that dimension. Each person will find herself placed at a different location along each dimension, and will have a personal sense of whether the legal system as a whole – that is, the minimum N in each dimension – represents a compromise that she can live with (is more or less compatible with her interests, given the extent to which she recognizes the need for the system to be also compatible with the interests of others). This is equivalent to the setup of the Hadfield/Weingast model, which supposes each citizen has an “idiosyncratic logic” that may converge to a greater or lesser extent with the “common logic” represented by the law.

Model a repeated ruler–population interaction as follows. Each round of the game, the ruler/ruling group sets a new policy along some dimension, and each citizen simultaneously chooses to send a costless signal: accept or reject. Then the ruler chooses whether to insist on the new policy or reset it to the status quo, and if the ruler insists on the new policy, each citizen chooses whether to resist (revolt, etc.) or not resist. If enough citizens resist, the resistance succeeds: the policy is reset to the status quo, and the ruler pays some positive cost. I submit that there is an equilibrium such that each citizen who thinks the legal system is generally in his interest sends the reject signal when the law is violated, and, for a sufficiently large number of citizens who think the law is in their interest, revolts if the ruler (off the equilibrium path) insists on the new policy.

First, some more intuition. Suppose the ruler violates the law along one dimension. Say, she enacts a law expropriating property from some disfavored religion and giving it to her allies in the majority religion. And suppose I am in the majority religion. Why might I want to nonetheless resist this enactment, even though it is in my interest? One reason is if I and the rest of the population are playing reciprocation strategies in which I can only rely on their support for resisting policies that hurt me if I support them in resisting policies that hurt them. If the legal system as a whole is more or less in my interest, such a reciprocation strategy may be sufficient to induce me to resist the illegal enactment. In other words, the reciprocal deal may lower my preference costs for any given policy by increasing the downside to me of the legal system’s failure. And I can credibly signal that the law is more or less in my interest by announcing my objections to the law, so long as that announcement causes me no costs other than the surrender of the religious preference.

Formalize this more-or-less-in-my-interest idea by saying that, as to each citizen, she meets the interests condition if the discounted value of the minimum guaranteed her by the law, g_i,L is greater than her discounted expected payoff in a situation in which the law fails and is replaced with lawless rule of the powerful in that society, E(g_i,A).

Now let us say that a group of people meets the overwhelming power condition if, should each member of that group who meets the interests condition resist the ruler, the probability of that group’s succeeding is sufficiently high that it is common knowledge among that group that, given the levels of, for example, risk aversion, disparate power distribution, and so on in that society, no member of that group as to whom the interests condition is satisfied would view it to be too costly to resist the ruler. This is a condition that resists full formalization, but is easy to approximate. In a society in which power is roughly evenly distributed, a moderate supermajority is likely to meet this condition; in situations with more disparate power, it may require correspondingly larger majorities (or the participation of the powerful). I need not specify the size of the group necessary to meet the condition in every (or any) society; rather, it need only be intuitively plausible that a possible group that can meet the condition exists in every society – and I submit that it is. At the limit, in any real-world society, the group “everybody but the ruler” (president, members of parliament, etc.) clearly meets the overwhelming power condition.⁶⁴

In slightly more formal terms, this condition rests on a model of rebellion such that each citizen controls a share Π of the overall distribution of power. These shares may vary (for example, military officers control more, small children control less), but I assume that the distribution is not so lopsided that a single official or small group of officials can hold off the rest – this merely excludes science-fictional societies in which a mad ruler controls a robot army by computer (and the undersigned dearly hopes that current drone technology does not progress to the point where this is possible in his lifetime, for it is difficult to see how the rule of law could survive in such a world). It is assumed that, if it can coordinate, the probability of a given side in a rebellion prevailing increases as the ratio of its aggregate power Πn_i to the aggregate power of the other side Πn_j increases, and is 1 at the limit; it is further assumed that as that ratio increases in favor of a rebellious party, the expected direct and retaliation cost E(C_r) of rebelling to each member of that party approaches 0 (if everyone rises up at once, they win quickly, cheaply, and painlessly). Then the overwhelming power condition describes any group that controls a sufficiently large share of power such that for each member of the group who meets the interests condition the impact of the risk of loss and the cost of rebellion on his or her decisions is negligible, or, letting Φ designate the probability of the given group winning a conflict: for all i in the group, if g_i,L > E(g_i,A), then Φ(g_i,L)+(1 – Φ)E(g_i,A) – E(C_r) > E(g_i,A).

Finally, I specify one more condition, relating to the history of play: the prior-consistency condition. In every previous round in which the ruler announced an illegal policy, the number of people who signaled rejection met the overwhelming power condition; and in each such previous round in which in addition (a) the ruler did not withdraw the policy change, and (b) the number of people who signaled met the overwhelming power condition, the number of people who actually resisted met the overwhelming power condition.

With that in hand, I claim that the following strategy set is in subgame perfect equilibrium in indefinitely repeated games, for sufficiently low discounting, and if a sufficiently large number of people meet the interests condition: if the prior-consistency condition is true, the ruler specifies his or her most preferred policy that complies with the law, and if the prior-consistency condition is false, the ruler specifies his or her most preferred policy; in either condition, each nonruler citizen always signals acceptance of policies that comply with the law, and acceptance of all policies if the citizen does not satisfy the interests condition. Each nonruler citizen who satisfies the interests condition always signals rejection of policies that do not comply with the law. If the policy complies with the law, the round ends and the policy is implemented. Otherwise:

1. (A) the ruler withdraws the policy change and each citizen who signaled rejection resists if the ruler does not withdraw the policy change, if and only if:

   1. (A-1) the number of people who signaled rejection meets the overwhelming power condition, and

   2. (A-2) the prior-consistency condition is true; otherwise

2. (B) the ruler insists on the policy change and no citizen resists.

In all cases, the behavior of citizens who do not signal rejection and those who do not meet the interests condition is irrelevant; I shall assume they never resist or signal rejection, respectively.