Meritocracy and the Tests of Virtue in Greek and Confucian Political Thought

A. Difficulties in Identifying Virtuous People

The problems of identifying people of genuine virtue were real, practical challenges for many Confucians, many of whom were responsible for or directly affected by the policies and practices by which a person’s qualifications for political office (including their virtue) were assessed. As is often the case for a vibrant tradition of political thought that takes some of its bearings from the actual work of governing, there is a great array of sometimes mutually incompatible arguments and ways of framing the problem of identifying genuine virtues in people. Nevertheless, as we will show, many Confucian political thinkers did cast their arguments and ways of framing the problem in terms set out by the Confucian classics, most of all by Confucius’s suggestive remarks about identifying people of true virtue in the Analects 論語 and by comments on character assessment in the Mencius (Mengzi) 孟子, and this puts us in a position to make some general (not universal or exceptionless) and defeasible claims about how Confucian political thinkers tended to understand the problem of identifying people of genuine virtue. In this subsection, we will discuss two problems that figure prominently in the historical texts: (1) the problem of seeing through flattery and superficial loyalty, and (2) the problem of discerning wholehearted virtue in the face of fakery. The first of these is already well known and has been theorized extensively by philosophers and intellectual historians who work on Chinese texts. The second is known and sometimes mentioned in recent scholarship, but there is little present-day discussion of the underlying psychological underpinnings and solutions to the problem, as the historical Confucians understood those underpinnings and solutions.

The first, well-known difficulty is that of seeing through flattery and superficial loyalty to make sound assessments of character. Roughly, it describes the tendency of people of all kinds to be quick to attribute virtues to people whose speech and behavior aligns with their superficial (often, immediate and present-minded) perception of their own best interest, and quick to attribute vices to people whose speech and behavior contravenes those interests. In the early Confucian texts, Confucius and Mencius warn their disciples to be wary of “village worthies” (xiangyuan 鄉原), who become popular among their contemporaries by according with the low standards of their time and place and excusing or flattering their fellow villagers for holding them (Analects 17.13, Mencius 7B.37). For rulers or other political authorities, the worry is that the political authorities will be too quick to attribute virtue to sycophants and advisors who agree with their policies and political decisions, while genuinely virtuous advisors are often critical and say things that the ruler doesn’t want to hear. The assumption that virtuous advice will often be unwelcome or offensive to the powerful is ubiquitous in classical representations of sage advisors (Henry 1987, 26). Generally speaking, truly virtuous advisors are represented as devoted to advancing the real interests of the people they advise (e.g., the long-term peace and stability of the ruler’s state) as opposed to superficial and present-minded interests. Accordingly, the ways to overcome this challenge are to care more about one’s real interests than one’s superficial and present-minded ones, and to welcome and cherish friends, teachers, and advisors who aren’t afraid to criticize (Henry 1987).

The second obstacle to correctly recognizing virtuous people has to do with faking or creating the pretense of virtue. In Bryan Van Norden’s influential study of virtue ethics in early Confucianism, he describes what he calls “counterfeit virtues”—“virtues” falsely attributed to people for behavior that is consistent with that of truly virtuous people but is actually done from insincere motives. Van Norden makes an example of a politician who pretends to have a good marriage to appear like a good family person in the eyes of ordinary voters (Van Norden 2007, 41–43). More broadly, we can imagine all manner of public officials who pretend to have compassion and respect for the people under their charge, but in fact only care about those people insofar as the perceived welfare of the people advances their own careers.

The challenge posed by counterfeit or faked virtues is quite different from the challenge of seeing through flattery and superficial loyalty. One might succeed at courting a friend who isn’t afraid to criticize and appears to maintain the high moral standards of the ancient Confucian sages, but who doesn’t sincerely or wholeheartedly uphold those standards and only pretends to be demanding or critical. We herewith offer our own reading of how Confucians who have worried about this problem understand its psychological underpinnings, and thus how they tend to characterize the solution to the problem. For ease of reference, we will call this the challenge of identifying sincere or wholehearted virtue.

The psychological picture that underlies the challenge is as follows. Sincere or wholehearted virtue is such that one cannot go about faking it all of the time. Faking virtue takes great effort and mental attention, both of which are finite resources, so that less-than-fully-virtuous agents will invariably need to let their guard down and reveal their true character. An important upshot is that you can see a person’s true character by observing how they behave when they are “off duty”—at leisure or at rest, for example. Here are two memorable and highly influential passages from the Analects that strongly suggest this view:

The Master said, “I can talk all day long with Yan Hui without him once disagreeing with me. In this way, he seems a bit stupid. And yet when we retire and I observe his private behavior, I see that it is in fact worthy to serve as an illustration of what I have taught. Hui is not stupid at all.”⁴ (Analects 2.9)

The Master said, “Look at the means that a person employs, observe the basis from which he acts, and discover where it is that he feels at ease. Where can he hide? Where can he hide?”⁵ (Analects 2.10)

There are, of course, many ways that one might explain the fact that a person’s true character can’t be hidden when they are off duty. In the interest of offering a clearer account of the underlying psychology of virtue that seems to be at work in much of Confucian virtue politics, we propose the following general schema. First, one of the ways in which genuine virtue can be distinguished from false virtue is by the indefatigability of the virtuous agent. When a truly virtuous person does virtuous things (e.g., shows compassion for the needy, teaches disciples, cares for her parents), she does it willingly, enthusiastically, and does not find it taxing or trying. As Confucius says in Analects 7.34, “What can be said about me is no more than this: I work at it without growing tired and encourage others without growing weary.” Second, the reason that people with false virtue can’t keep up the pretense of virtuous behavior all day is because they lack another constituent of true virtue, which we characterize as wholeheartedness. Roughly, wholehearted virtue is the characteristic of having well-integrated psychological dispositions, such that there is no internal friction or resistance to virtuous behavior. A person who visits a sick parent in the hospital willingly and out of a felt desire to comfort the parent is more wholeheartedly filial than someone who visits the sick parent reluctantly or grudgingly. Someone who quits an egregiously corrupt employer without hesitation is more wholeheartedly righteous than someone who stalls or spends some time looking for justifications to keep the position. The wholeheartedness criterion figures prominently in classical Confucian texts, helping to explain—for example—why Mencius thinks that even nonconscious processes should accord with one’s explicit, virtuous aims, and why Confucius regards “taking joy in the Way” as better than merely “understanding the Way” or “loving the Way” (Mencius 4A.27; Analects 6.20). The view that wholeheartedness is a necessary feature of full virtue became an explicit part of Confucian virtue theory in the early Daoxue 道學 or Neo-Confucian period of virtue theory, when many Confucians endorsed the view that genuine or complete virtue must be “sincere” (cheng 誠) and regularly characterized sincerity as marked by a lack of internal division or self-deception (ziqi 自欺) in one’s own heart-mind (Great Learning [Daxue 大學], commentary 6; Zheng 2022, 21–50; An 2005, 144–50; Tiwald 2018, 179–83; 2023, 81–88).

Historically, in the long tradition of Confucian political thought that followed Confucius and Mencius, the prevailing conclusion drawn from this picture is that one can reliably determine a person’s character through close observation over time. The assumption is that most people can “fake” or create the pretense of virtue for some time—say, long enough to give a speech, to perform a public ritual, or to produce a high-minded essay for an examiner. But for those who fake virtue, unlike for those who are wholehearted, their vigor for virtuous activity will flag, and it is in a person’s unguarded moments—when engaged in private activities or at leisure—that one can discern a person’s true sources of motivation and what truly gives them comfort and joy. We offer this not only as a plausible reading of the two Analects passages taken on their own terms, but also as the best framework through which to interpret later Confucianism’s views about the difficult art of assessing someone’s true character.

To illustrate its prevalence in post-classical Confucian political thought, we offer the views of two figures who had an outsize influence on Confucian institutions and Confucian philosophy. The first is Lu Zhi 陸贄 (754–805), the reform-minded chancellor and political advisor whose writings came to represent Confucian political ideals in the Tang dynasty. In a petition to his emperor (Dezong 德宗, r. 779–805), he argued for a method of selecting political officials that relied on the recommendations of those who knew them best (usually their immediate or near-immediate superiors) as opposed to testing for virtue through civil service exams or centralizing recommendations in the hands of more distant and powerful officials:

It is only when a person has befriended someone for a long time, knowing comprehensively both his deep and underlying as well as his more evident and manifest sides, having a good grip on his sternest will and intention, seeing thoroughly his capacity for practical matters, can one be certain that he really has kept the Way and preserved his talents for future use. And only under such condition will crafty pretenders find no room for their fakery [wei 偽]. Thus Confucius said: “Look at the means that a person employs, observe the basis from which he acts, and discover where it is that he feels at ease. Where can he hide?” [Analects 2.10]. Adequate observation certainly cannot be done within a single day and night.⁶

Notice in particular that Lu assumes one can only keep up fakery or pretense for a certain duration, such that true character will invariably show itself to careful observers over a long period of time.

Another giant of post-classical Confucianism is Zhu Xi 朱熹 (1130–1200), who also sees the strain of faking virtue as the key to explaining both Analects 2.10 and the need to discern the sincerity of one’s virtue by careful observation over time. In discussions with students, Zhu glosses the phrase “looking at the means a person employs” as “observing how the person is in his daily, ordinary life” (guan ren zhi fanri 觀人之凡日) (Classified Sayings of Master Zhu 朱子語類, juan 24, section 34). An astute student offers the following summary of his teacher’s views on the stepwise process of discerning someone’s true character:

Look at whether a given person does good things or bad things. If they do good things, you next need to observe where their motivation [yi 意] comes from. If the root motivation is something they should do for the sake of improving themselves and they do it effortlessly, then it is indeed for the sake of self improvement. If they do it thinking that they can be recognized for their good deeds by others, then the place it comes from is not good. If the place it comes from is good, then you must discover whether the person’s inner heart-mind is made joyful or not. If the inner heart-mind joyfully engages in good behavior, then they won’t be inclined to detest it or regard it as tiring, and so they will experience progress daily. If the inner heart-mind lacks that which gives it joy, then whether they do the task or abandon the task before completion, they will be unable to avoid faking it [wei 偽]. If one examines a person in this way, one will catch little, incremental glimpses of the secret and subtle designs of their heart-mind; this is a most difficult task. It’s also necessary that one be able to understand doctrines⁷ and fully grasp the Patterns [li 理] in oneself, so that one’s heart-mind can fully and comprehensively penetrate the Way and one can carefully and skillfully discriminate between what’s so and what’s not. Only then can one discover whether a person is like a sage.⁸ (Classified Sayings of Master Zhu, juan 24, section 23)

In Zhu’s reply, he approves of this account, adds the reason to examine the person for signs of joy is to determine whether they do good sincerely (cheng 誠), and reminds his audience that sincere virtue is unforced (Classified Sayings of Master Zhu, juan 24, section 23, in Zhu 1983).

In short, Zhu Xi shares the view that we find in Lu Zhi and in the Analects itself, according to which one assesses the virtue of someone through close observation over long periods of time, preferably including time when the person observed is “off duty” (at leisure or engaged in daily activities). This is because truly virtuous agents are wholehearted (“sincere”) in their virtuous activity, and thus don’t need to force themselves at the price of exhaustion, inner friction, and dissatisfaction.⁹ Both Lu and Zhu found support for this psychological picture of sincere virtue in Analects 2.10.

B. Local Recommendations

Today, when sinologists and members of the political commentariat talk about China’s meritocratic history, they often have in mind the policy of making political offices eligible only to candidates who earned high degrees by performing well on civil service exams, which consisted of significant amounts of essay-style writing on the Confucian classics and histories. To set the historical record straight, however, there is no proposal in the classical Confucian texts to select political officials by examination, and furthermore, the widespread practice of selecting political officials by examination is a relative latecomer in the history of Chinese meritocracy. The first nationwide exam for civil service is usually said to have been established by Emperor Yang of the Sui dynasty 隋煬帝 in 605 CE, but it was not until the Song dynasty (960–1279) that it became the primary means of appointment to political office (Kuhn 2009, 121–22). These historical facts alone do not suffice to show that selection-by-exam is un-Confucian, but it does put the onus on Confucian defenders of meritocratic exams to show that such a method could be made compatible with core Confucian principles.

There is a stronger textual and historical case for a method that Confucian writers called “selecting and promoting officials by recommendation from locals” (xiangli juxuan 鄉里舉選 or xiangju lixuan 鄉舉里選), which we will abbreviate as the method of “local recommendations.” According to this method, the person responsible for making appointments to high office (the ruler or a designee) gathers information about potential candidates from reliable people who know them best, largely construed as local officials and elders. The ruler or other central authority still makes the decision about whom to appoint or promote, but insofar as the decision depends on assessment of the candidate’s character, they defer to local judgments of the candidate. Of course, it didn’t always work this way, and there were many periods in the history of Confucian regimes when appointments and promotions were more or less at the whim of an office or official in the ruler’s court. But insofar as we want a method that influential Confucians have regarded as a desirable ideal, the evidence for a system of local recommendations is much stronger than the evidence for civil service exams.

We offer three brief arguments to show this. First, if we look at the Confucian canon, we find more passages that suggest or point to this method as better warranted by the classical Confucian account of virtue. We have already discussed Analects 2.9 and 2.10, which strongly imply that close and extensive observation of a person while off duty is necessary. Another important classical source is Mencius 5A.5, in which Mencius endorses the method by which the legendary sage-king Yao selected the legendary sage-king Shun (who wasn’t his own biological son) to succeed him. Mencius says that Yao gave Shun a long trial period in various lower-level managerial positions, and then waited for signs from both Heaven and the people under his supervision that he was worthy of the highest promotion. And then there is a muchdiscussed proposal in Mencius 1B.7, where Mencius suggests that certain political actions that require high-stakes assessments of a person’s “worthiness” (xian 賢) should be resolved through a combination of personal examination and deference to “the people of the state” (guoren 國人). Mencius gives as examples the decision to demote a bad minister, execute someone accused of a crime, or promote someone to high office—in all of these cases, the ruler should await an overwhelming consensus on the part of the people of the state before examining the person himself.¹⁰

A second argument begins with the historical observation that a system of appointment and promotion by recommendation predates the civil service examination system by many centuries, and remained strong even in Sui and Tang dynasties, whose governments administered exams but continued to make the vast majority of appointments by recommendation. To be sure, the recommendation system did not always give weight to the recommendations of local leaders or peoples—sometimes it was just bent to the will of the ruler or a powerful minister. But it nevertheless remained a desirable ideal to delegate the work of assessing a candidate’s character to the people who knew the candidate best. Lu Zhi’s aforementioned petition, for example, was an impassioned plea to restore the more traditional system of “selecting and promoting by recommendation from locals” and take that power away from a central authority (Lu n.d.).¹¹ Returning to the old local recommendation system became a perennial cause of reform-minded Confucians during the heyday of the civil service exams (Nivison 1963; Chaffee 1985, 58–59). Insofar as actual institutions in Confucian China often relied primarily on character assessments made by central authorities rather than locals, we suggest that these are better understood as departures from a Confucian ideal for the sake of expediency or partisan advantage rather than an ideological move grounded in Confucian principles.

A third reason to treat the system of local recommendations as a more central feature of Confucian meritocracy than civil service exams is because it remained an important factor even in the (relatively late) periods when the exams became the primary pathway to officialdom. Consider how the civil service exam system was meant to work: both in fact and in principle, it provided a credential for a certain range of important offices, but appointments were never made on the basis of exam results alone. Earning a degree through the examination system was a necessary but not sufficient condition for an appointment as chancellor, censor, regional governor, etc. As historians have often noted, for much of the Song through the Qing dynasty—the supposed heyday of exam-based meritocracy—the number of eligible degree-holders vastly outnumbered the number of significant managerial positions (for example, see Kuhn 2009, 123–24). Other important considerations were necessarily factored in when making appointments, and recommendations continued to be very influential (Xiao and Li 2013, 347–48). Even in the Song dynasty, various guarantees of good character were required. Most of these guarantees were perfunctory (for example, families had to submit and then get an official to verify biographical data), but there was often an aspiration in the central government to strengthen the procedures so as to require a deeper investigation into a candidate’s background. Such reforms were adopted, for example, in 1000 and 1026, and some of the reports on candidate backgrounds are still archived (cases where criminal activity or fabrication of biographical data was detected). But it seems that these procedures proved costly and time-consuming, and it is not clear how long they lasted (Chaffee 1985, 58–61). What this does show, however, is that the method of relying on locals to assess a candidate’s character continued to be regarded as an important one by Confucian lights, even if not one that was always practicable or feasible to implement.

While the method of local recommendations is more faithful to the traditional Confucian account of virtue and the recognition of virtue, it nevertheless faces some profound problems at the level of political implementation, particularly in large-scale states or nations. These problems are well attested both by history and by critics of “elevating the worthy” (shangxian 尚賢) in early political discourse. As Yuri Pines points out, there are certain problems characteristic of any system of appointments that depends on recommendations (local or otherwise). First and foremost, and as articulated powerfully in the Legalist classics the Book of Lord Shang (Shangjun shu 商君書) and the Han Feizi (韓非子), there are various problems that arise from the fact that virtue lacks “objective criteria,” so that most rulers had to rely on highly subjective impressions of a candidate’s reputation, which is far more likely to reflect partisan loyalties or factional preferences than reliable judgments of character. This, in turn, led many of the great Legalist thinkers to recommend selection and promotion on the basis of more objective criteria, such as the number of enemies killed in battle or units of food produced on one’s farmland (Pines 2013, 181–84; Book of Lord Shang 商君書, ch. 6., in Liang 2017 and Pines 2017). To state the problem a bit more carefully: discerning the quality of a person’s character is such that a recommender’s judgment cannot be easily confirmed or disconfirmed by public authorities, so that public authorities have to trust the recommender without recourse to an independent basis of judgment. One problem is that public authorities do not generally have the rare combination of wisdom and good character that would make them a keen judge of virtue in others. Another problem is that discerning true virtue requires enormous amounts of time and effort: Confucius implies that one must live together and interact daily for months or years, as he did with Yan Hui. No political authority presiding over a large state will have the time to confirm or disconfirm all of the recommendations made to them, and so it is unavoidable that the ruler would have to defer to the judgments of others. So the real problem is that political authorities have no independent means of easily confirming the recommender’s judgment of a candidate’s character. Without an independent means of easily confirming (or disconfirming), they are left at the mercy of the local leaders who make the recommendations, and local leaders are much more likely to recommend on the basis of factional considerations than on sound assessments of character.

However we may want to characterize the Legalist worry about factionalism, their worry was borne out by historical events. From the late Han through the early Sui dynasties, selection and promotion for high office was based largely on local recommendations. Recommendations often were explicitly based on the good character of the candidate. Good character would presumably be well distributed across a local population, and yet offices were generally controlled and occupied by members of a small set of powerful families. The system of “local ranks” (xiangpin 鄉品) were notoriously entrenched and stable from one generation to the next, and there is little doubt that family affiliation and having the right connections were the primary factor in determining most appointments (see Ebrey 1990 and Graffin 1990). As Pines points out, this strongly suggests that the method of local recommendations, in its various historical iterations from the late Han to the early Sui, was both an obstacle to class mobility and, most importantly, bad at selecting for virtuous officials (Pines 2013, 187–88).

C. The Examination System

Examinations played at least a small part in the selection process for some offices as early as the Eastern Han dynasty (25–220 CE), however, they were one requirement among others—including the requirement of strong recommendations—for a subset of offices that were thought to require special sorts of knowledge or talent, such as flood control. The earliest iteration was meant primarily to test for talent or a kind of technical know-how and not moral quality or virtue (Xiao and Li 2013, 341–42). Moreover, the number of offices for which a successful exam result was required was very small. The first significant nationwide civil service examination system was set up in the Sui dynasty, primarily as a way for the emperor and his court to curb the power of aristocratic families that dominated the local recommendation system. This practice continued in the Tang dynasty, but in neither dynasty did it become the primary pathway to high office. As late as the Tang dynasty (618–907), over 90 percent of state officials were still getting their appointments by recommendation and not through civil service exams. It was finally in the later Song dynasty that exams became the primary vehicle to high office, and the number of candidates sitting for the exams ballooned from “hundreds” to “hundreds of thousands” (Kuhn 2009, 121–22). The contents and criteria of evaluation took many different shapes over the generations, with some iterations (such as in the Tang dynasty) emphasizing the ability to compose poetry and others (such as those implemented with Wang Anshi’s New Policies reforms in 1104–1121) emphasizing technical and specialized knowledge (see Xiao and Li 2013, 344; and Xiao 1982, 491–92).¹² But some mastery of the Confucian classics was required for almost every iteration of the exams. When Zhu Xi’s brand of Confucianism became state orthodoxy in 1313–15, the emphasis shifted decisively toward knowledge of the more philosophical Confucian classics known as the Four Books, as opposed to histories, poetry, or public policy issues. The Four Books continued to dominate both the exams and the weighting of grades for most of the late imperial period (see Elman 2020, 415–25).¹³

Once civil service examinations became the only accessible pathway to China’s most prestigious jobs—and means of entry into China’s most prestigious class—they also became a topic of great public interest and debate. Most of the major Confucian philosophers of the Song dynasty and thereafter were critical of both the civil service exams as designed and implemented in their day and of the way that fierce competition and the desire for upward mobility warped the otherwise healthier temperaments of candidates and their families. These include highly influential thinkers like Lu Zhi (754–805), Wang Anshi (1021–1086), Cheng Yi (1033–1107), Zhu Xi (1130–1200), and Wang Tingxiang (1474–1544). Their shared assumption was that exams tended to test various skills and abilities that were not well correlated with virtue, such as the ability to write elegantly (Xiao and Li 2013, 345–46). Lu Zhi did not propose abolishing civil service exams, but thought that relying on exams alone—without the further verification of the local recommendation system—would reward candidates for “crafty fakery” or hypocrisy (qiaowei 巧偽) (Lu n.d.; see also Chiu-Duke 2000, 152).¹⁴

Among the many criticisms these philosophers raised, Zhu Xi’s chief complaints were that the exam system created perverse incentives to impress the examination officials (graders or assessors) with novel or original interpretations, so that eccentric answers tended to win the candidate higher esteem than faithfully reporting the original meaning of the classics (Classified Sayings of Master Zhu, juan 111, sections 4 and 13). In one discussion with his students, he compares the demand to write novel exam answers to pressing used grapes:

These days, in the exam essays on policy and strategy, those who tend most to be regarded as virtuous write even more unreadable essays. Because there are only so many current issues and all of them have already been discussed, naturally they can’t say more about those. It’s like pressing grapes to make wine. The first time you press them they make juice for wine. The second time you press them they make some more and the third time they make more still. If, like today, they keep going back to so many dregs and working on them to make more wine, what will they have to say? Since there is nothing to be said, they only concern themselves with what’s new and original. The most damaging part is that they treat virtue lightly, destroy reputations and integrity, worship cleverness in tactics, and esteem what’s surprising and deceptive. Reading these essays pains one’s heart and hurts one’s head.¹⁵ (Classified Sayings of Master Zhu, juan 111, section 34)

Perhaps Zhu was especially troubled by the built-in incentives to impress bored examiners because he knew from personal experience how little actual understanding was required to elicit a positive response. In one of his more striking autobiographical confessions, Zhu recounts meetings with a Chan monk who taught him how to give evasive, nonsensical (hushuo 胡說) answers to deep questions, and admits to using the monk’s ideas to successfully pass his own civil service exam at 19, despite his own poor grasp of the ideas (juan 106, section 38). At the broadest level of description, Zhu’s worry was that test-takers were encouraged to prize various skills and achievements more highly than having an accurate understanding of what’s actually the case. These competing values included novelty, cleverness, style, and avoiding offense (juan 111, sections 25 and 32).

Zhu criticized other aspects of the examination system of his day. For example, many candidates were prepared for the exams by government-run schools that appointed their instructors primarily on the basis of seniority rather than knowledge or scholarship. Zhu also strenuously objected to the outdated system of regional quotas which determined how many candidates could be sent from each prefecture (juan 111, sections 20 and 22). These were criticisms of the system in place at his time, and not necessarily objections to using civil service exams to select for virtue as such; so did Zhu think that the civil service exams could be improved so that they would play a meaningful role in identifying virtuous candidates for high office?

Zhu’s views on this more fundamental question are a little difficult to discern. As other modern scholarship has noted, Zhu wrote a proposal for reforming the examination system, one that would give candidates sufficient advance notice to prepare the texts on which they would be tested, would more fairly distribute seats across the prefectures, and would test candidates on something we might describe as their powers of critical thinking (Nivison 1963; Classified Sayings of Master Zhu, juan 111, sections 22–24). Given his interest in reforming the exams and the system by which it was administered, one might think that he saw exams as a reasonable way to assess the virtue of candidates for political office. However Zhu was enough of a political and institutional realist to see that the exam system was likely to endure for a long time, good or bad. Given this, he thought that at least the examination system should be fair, should be supplemented with other methods of identifying candidates of good character, and should play a reasonably positive or constructive role in promoting knowledge of Confucianism and Confucian virtues. When a student asked whether the defects of the system warranted immediate action, Zhu replied that it just wasn’t possible to abolish it and proposed instead that it should be reorganized according to a clearer and more coherent set of principles (Classified Sayings of Master Zhu, juan 111, section 17). In other discussions with students, Zhu made clear that the primary work of selecting for virtue should be done by those who are well acquainted with the candidates for office, and suggested that the selection of candidates to sit for the exams should be done by the old method of local recommendations (xiangju lixuan) (juan 111, sections 3 and 19). But other comments suggest that he saw any reinstatement of local recommendations as unviable and impractical in the political and institutional culture of his era (juan 111, section 15). At the very least, he thought the political culture too corrupt to produce reliable and faithful reports from local leaders (juan 111, section 15). Perhaps the recent, cumbersome attempts to conduct more investigation into the backgrounds of candidates (initiated in 1000 and 1026 and discussed above) showed him that the system was too costly and time-consuming for the state bureaucracy of the Southern Song. Despite his interest in reforming the examination system, then, it seems relatively safe to say that he was under no illusions about the inability of exams to test for virtue. If anything, Zhu’s case shows that he largely took for granted what seems to be a widespread assumption that assessments of a person’s character can only be made by those who know them well, which seems to motivate his suggestion that, ideally, candidates for the exam would be selected by the older system of local recommendations. If this is right, then Zhu Xi’s views further substantiate the claim that the method of local recommendations and not test-taking is much closer to the heart of the Confucian philosophical and political vision of selecting virtuous officials.¹⁶

A. Aristotle

Despite being one of the most famous proponents of rule by the virtuous, Aristotle gives us very little guidance on how to structure institutions so that virtuous people hold office. His general approach is to point out that democratic constitutions are more inclusive in terms of who participates in rule, emphasizing the equality of the (male) citizens in their status as free, whereas oligarchic constitutions are less inclusive, emphasizing the inequality of citizens on the basis of wealth, education, or family lineage (Politics III.9, 1280a7–25). On this analysis, aristocratic constitutions (understood as rule by the best) are more similar to oligarchic constitutions insofar as they emphasize the inequalities of citizens, but make distinctions on the basis of virtue.

But, perhaps surprisingly, Aristotle does not give us descriptions of how to get the virtuous in office rather than the wealthy or wellborn. Instead, what we find is the more general point that democratic constitutions tend to favor selection mechanisms that do not discriminate between eligible candidates, namely by using random lottery to appoint officials, to those that do discriminate between eligible candidates through the use of election (Pol. II.6, 1266a5–26; II.11, 1273a13–30; II.12, 1273b40–41; IV.5, 1292b1–4; IV.7, 1293b7–21; IV.9, 1294b6–13; IV.9, 1294b29–34; IV.14, 1298b5–11; IV.15, 1300a8–1300b5). The basic idea is that if anybody can do the job, then random lottery is appropriate, but if the job requires skill or a certain kind of person, then election is appropriate. Thus, we can think of Aristotle’s constitutional design as having two stages: first, we ask if the position requires discriminating between eligible candidates; then, we ask on what basis we should make distinctions. Oligarchs and aristocrats will claim that many offices require particular qualifications, but aristocrats will distinguish at the second stage on the basis of virtue rather than wealth or lineage.

And that’s about all the explicit help we get: if you want an Aristotelian aristocracy, make the offices appointed by election, and make it clear that people should vote on the basis of the perceived excellence of the candidates. Alas, that’s not much help if we’re wanting a fresh meritocratic selection mechanism that differs from what contemporary representative democracies use.

Did Aristotle think, then, that election reliably selected virtuous people? He was surely aware of arguments that people do not vote on the basis of perceived virtue, because Plato in the Gorgias gives an extended criticism of citizen motivations to vote and Aristotle is well aware of that text. There, Plato claims that people vote for candidates who reflect their own values and who pander to their desires, not for candidates who genuinely improve the polis and its citizens (especially Gorgias 463a–466a, 510a–e, 512e–513c. See also Moss 2007; Kamtekar 2005; and Reid forthcoming). But Aristotle seems not to respond to this argument in our extant texts—perhaps because he thinks there is no better alternative, or perhaps because at some point public norms and education have to do the work in shaping the values on the basis of which people vote.

It is possible, however, to reconstruct plausible criticisms of putatively aristocratic institutions on the basis of Aristotle’s discussion. In particular, Aristotle seems to recognize that in practice meritocratic criteria quickly become bound up with adjacent oligarchic criteria (where “oligarchy” is understood as rule by the wealthy). For example, Aristotle says that citizens need leisure in order to rule well, and that citizens need wealth in order to have leisure (Pol. II.11, 1273a21–25). So wealth becomes a necessary condition of merit, but then people start focusing more on the wealth than the merit. Considering a Carthaginian law, Aristotle writes that, “even if one must look to wealth, too, in order to ensure leisure, still it is bad that that the most important offices, those of king and general, should be for sale. For this law gives more esteem to wealth than to virtue, and makes the entire city-state love money” (II.11, 1273a35–39). So the Carthaginians are right to ensure that their rulers have adequate leisure, but they began to mistake a necessary condition (wealth) for the goal (virtue).

Aristotle alludes to a similar concern in pointing out that aristocratic characteristics and oligarchic characteristics tend to come as a package, such that virtue is closely associated with wealth, good birth, and education (IV.8, 1293b36–42; IV.8, 1294a20–22; V.7, 1307a26–27.). Of course Aristotle is well aware that wealth and virtue are not reliably correlated, and that illustrious lineage is no guarantee of character (IV.11, 1295b1–1296b12). But we can see the problem: it is very hard to select for virtue directly, so we have to rely on imperfect but verifiable proxies like education, wealth, and family (not to mention gender or status as free). This makes aristocracies vulnerable to elite capture and shows how easily aristocracies can degrade into oligarchies, as selecting for virtue often amounts to selecting for wealth in practice.

We can muster further Aristotelian worries about meritocracy on the basis of other discussions in the text. In the first place, Aristotle is very concerned about the limitations on good government generated by large populations. “Now that city-states have become even larger,” he writes, “it is perhaps no longer easy for any other constitution to arise besides democracy” (III.15, 1286b20–22). Aristotle thinks that constitutions that do not allow large parts of the population a share in governance are unstable, and democracies most easily allow large numbers of people to participate in politics (V.1, 1301a37–39; III.11, 1281b25–34). But this is a bad outcome for Aristotle. Aristotle thinks that a population of 100,000 citizens is too big to constitute a single polis, and in describing his ideal state, Aristotle is careful to limit the size of the city because a large city could not be well organized or properly governed (EN IX.10, 1170b31-32; Pol. VII.4, 1326a25–1326b7). In addition, Aristotle worries that people will not know whom they’re voting for in a large city. In Politics VII.4, Aristotle writes:

For a city-state’s actions are either those of the rulers or those of the ruled. And a ruler’s task is to issue orders and judge. But in order to decide lawsuits and distribute offices on the basis of merit, each citizen must know what sorts of people the other citizens are. For where they do not know this, the business of electing officials and deciding lawsuits must go badly, since to act haphazardly is unjust in both these proceedings. But this is plainly what occurs in an overly populated city-state. (1326b12–20)

This passage makes explicit that elections and judicial procedures go badly when citizens lack knowledge of each other’s characters, and that this is a predictable result when the city-state becomes too large. For a further illustration of how population size affects the quality of governance, we can turn to Plutarch’s account of the famous ostracism (a ten-year imposed exile) of Aristides the Just in the early fifth century BCE: when the vote was being taken, Aristides went to the marketplace where an illiterate man in the crowd (not recognizing to whom he was talking) asked Aristides to help him write on the tablet to cast his vote in favor of ostracism. Aristides asked the man what harm he had suffered from Aristides, and the man replied, “None at all! I don’t know the man. But it’s annoying to hear him called ‘the Just’ everywhere.” Aristides then wrote his own name on the tablet, handed it back to the man, and was exiled by popular vote (Plutarch, Life of Aristides VII.5–6).

More generally, Aristotle is explicit that virtue is rare and it is difficult to get to know somebody’s character. In the Politics, Aristotle writes that, “good birth and virtue are found in few people, whereas wealth and freedom are more widespread. For no city-state has a hundred good and well-born men, but there are rich ones in many places” (Pol. V.1, 1301b39–1302a2; see also V.4, 1304b2–5). In addition, even with our friends Aristotle recognizes that it takes a very long time to get to know somebody’s character and requires sharing activities with them (EN VIII.3, 1156b25–29). This aspect of Aristotle’s ethics perhaps explains why Aristotle himself does not detail more meritocratic institutions even though it is clear he endorses the ideal: virtuous people are rare and hard to identify even in close interpersonal settings, let alone at the level of the polis.¹⁸

Finally, Aristotle would be extremely skeptical of using an examination system as a test for virtue. A central tenet of Aristotle’s psychology is that theoretical and practical reason are different faculties (EN VI.1, 1138b35–1139a17). Aristotle is scathing of those who say the right things and can make all the right arguments but whose desires and actions do not harmonize with reason (I.3, 1095a2–11; I.4, 1095a30–b13; X.9, 1179a35–b31). Examining people on their understanding of ethics amounts to testing their theoretical reasoning, but the kind of virtue we need for governance would require testing character virtue and practical reasoning—but this can only be done by observing how somebody lives, not by seeing what they say.¹⁹ The problem is exacerbated when it is common knowledge what people are supposed to say, whether or not they actually believe it. So the examination system is at best a bad proxy for the thing we want to track.

Aristotle’s main suggestion for how to implement meritocratic institutions, then, is to rely on election. Insofar as the election of representatives is already a core feature of most democratic institutions, and (at least according to the Federalist Papers) is meritocratic (Macedo 2013; see e.g. Federalist #62, 63, 64, 68, 71, 72, 73, 76),²⁰ Aristotle does not offer us any new mechanisms for selecting the virtuous and in fact gives us many reasons for thinking that appointing on the basis of virtue will be very difficult.

B. Plato

Plato’s Republic proposes a constitution where philosophers will be rulers, and “philosophers” is understood to mean people who have attained the heights of virtue and political skill. So how did Plato envisage that we would train and identify the people who had the potential to scale these heights?

The philosopher’s education is a lifelong enterprise, taking at minimum 55 years of training, where the potential philosopher-rulers are tested repeatedly and observed closely, including 15 years serving in minor military and civic offices.²¹ Plato’s education system in the Republic is a series of increasingly difficult tests, both intellectual and practical, where those who fail to meet the standards do not get the opportunity to progress to the next stage. What’s important is that whoever assesses the citizens has decades of close personal observation to work from. Furthermore, Plato’s concern is not that there will be too many qualified people to decide between, but that there will not be enough people who pass all of the tests.²² This means that the examiners could focus on a small group of promising candidates.

To put it mildly, some have worried about the feasibility of Plato’s proposals in the Republic. One important concern is how things would work in the first generation: if only those who have been raised in the ideal city will develop the appropriate characters to run the city (and sort the wheat from the chaff in the next generation), then how will the education system get started (Annas 1981, 186–8)? Even if we grant that virtuous people could identify other virtuous people or those with the potential for virtue, we still have to find the first group of virtuous people to run the show.

In addition, Plato has a doubly peculiar conception of virtue. Drawing from his own cultural context, he underscores the significance of excelling in music and gymnastics as a propaedeutic to virtue (a viewpoint shared by Aristotle). However, Plato uniquely assigns substantial weight to abstract reasoning in mathematics, science, and logic for attaining knowledge of goodness. It is worth noting that if we are concerned about assessing unreliable proxies for virtue, Plato may face a strong objection here. One's proficiency in music, gymnastics, and mathematics does not reliably indicate virtuous character.

Although the Republic is notoriously obscure on many details of constitutional design, fortunately Plato’s Laws gives us much more guidance.²³ Office-holders in the Laws are appointed through a series of complicated rounds of nomination, election, and random lottery to select the final candidates.²⁴ Though we are not given the justification explicitly, Plato seems to be relying more extensively on election for offices that require skill and virtue to rule well, and only uses lottery sparingly for offices where he can afford to encourage public participation in politics without compromising good governance. So we see once more that election is the preferred mechanism for selecting the meritorious.

Plato also ensures that the city of the Laws is kept at a manageable size such that citizens can know the people they are nominating and voting for. The population of the city is limited to 5,040 households, and divided into 12 districts, so presumably citizens would know those in their own district especially well.²⁵ An important aspect of civil society in the Laws is regular sharing in eating and drinking. Plato institutes Spartan-style shared mess halls for the various districts, and Athenian-style drinking parties (symposia), which he claims would also serve an important function in testing character, as he thinks that deep aspects of our character and values are revealed when we are drunk (648a–650b. See also Baima 2017, 2018; Landauer 2022). Again, we might worry about the effectiveness of this particular test of virtue, but what is worth highlighting is the communal living and thus the close acquaintance of fellow-citizens that is encouraged in the Laws. Fruitful comparisons can also be made between Plato’s use of alcohol to reveal character and Confucian recommendations to observe people when they are off-duty and when their guard has been let down.

What about the concern from the Gorgias that citizens will vote for reasons other than the merit of the candidate? Well, strictly speaking, Plato’s worry was that public oratory exacerbates bad voting practices, and there appears to be no profession of public oratory in the Laws.²⁶ So perhaps the view here is that election is the best mechanism we have to select for the skilled and virtuous, but we should be careful to curtail the forces that could undermine the quality of this mechanism. Given the parallels between Greek rhetoric and contemporary marketing or public relations, it is plausible that Plato would advise us to focus less on finding an alternative to election and to focus more on the forces that undermine the integrity of our elections and that give us bad information about the people for whom we are voting.

There is one final insight from the Laws that is worth drawing attention to. For Plato, it is the auditors even more than the rulers who have to have the most complete virtue (945e2–3). Plato relies heavily on the standard Athenian practice of submitting officers to a scrutiny before holding office and audit after their time in office is complete (Annas 2017, 35–36). But Plato and Aristotle disagree about who should serve this function. Aristotle recommends the Athenian practice of giving the public oversight over the performance of office-holders (Pol. III.11, 1282a7–41; Lane 2016), whereas Plato makes this institution meritocratic. Both positions are defensible, but have their costs. Aristotle’s view rests on the idea that the users of the product are often better judges than the makers, so he thinks that the public should be arbiter of the quality of laws, because they have to live under them. But Plato consistently points out that public doesn’t have the relevant knowledge to know what the standards of good legislation are, so he thinks that only virtuous political experts are in an epistemic position to judge the quality of laws (Gorgias 464d–465a; Statesman 298a–300a; Rep. 487d3–489c6; see especially Trivigno 2021).²⁷ Aristotle concedes that the user/maker argument is limited in scope, so he grants the Platonic criticism in some domains (Pol. III.11, 1281b38–1281a14); but Plato’s view risks insulating the office-holders from valuable feedback from the public and increases the risk of rulers abusing their power.

The reason this discussion is important is that it highlights the three domains in which virtue may need to operate: (1) the rulers themselves need to be virtuous; (2) those selecting the rulers need to be able to identify the virtuous; (3) those auditing the rulers need to be able to assess reliably the performance of the rulers. It may be that it is possible to perform the tasks of these latter two functions (selection and oversight) without being virtuous, but there is at least a major concern that there could be a mismatch between standards being applied in each of these domains. Meritocrats could argue that this problem is overcome by ensuring that those involved in selection and oversight are virtuous too, but if we were already worried about the scarcity of virtuous people for filling offices of rule, then we should be very worried about a lack of adequate candidates for selection and oversight.

For the sake of completeness, there is one last meritocratic suggestion in the Platonic corpus worth mentioning, which is that the Statesman suggests the possibility of a knowledgeable and skilled statesman advising a ruler rather than ruling himself (292d–293a, see also Laws 710c–d). This idea has many affinities to the Confucian idea of scholar-advisors serving the Emperor, but arguably this has already been implemented in a contemporary setting in the form of special advisory committees and non-elected branches of the government administration (see Macedo 2013). While these aspects of modern bureaucracies are not as glamorous as an explicitly meritocratic house in the legislature, we should not underestimate the extent to which meritocratic institutions have already been incorporated into our government systems.

So Plato and Aristotle offer us no fresh or new meritocratic selection mechanisms that are likely to work, and they draw our attention to the importance of close personal observation of character over long periods of time. Given their concerns about how tests of virtue can become tests for bad proxies of virtue (or about something else entirely), it may be that the best way forward for meritocrats is to try to improve existing meritocratic institutions rather than search for radical new alternatives.