1 Introduction

A lively debate has been playing out over the last decade or two on whether various forms of online friendships pass muster as instantiations of ‘character friendships’ in the Aristotelian sense. As often in philosophy, there is a formidable anti-camp (see e.g. Cocking and Matthews 2000; Vallor 2012; Fröding and Peterson 2012; McFall 2012; Sharp 2012) but also a formidable pro-camp (see e.g. Elder 2014; Kaliarnta 2016; Bülow and Felix 2016), with both offering palatable and initially persuasive arguments. It could be argued that this debate has reached something of a saturation point and that expending further intellectual energy on it would be redundant. While my sympathies happen to lie with the pro-camp, as becomes evident in due course, my rehearsal of the general terms of the existing debate will be reduced to a stage-setting critical overview.

Much of the relevant debate used to focus on friendships forged on social-media platforms such as Facebook (see e.g. Deresiewicz 2009). The rapid developments and fiendish intricacies of those platforms, especially in light of their commercial nature and potential abuse, muddy the intellectual waters considerably when delving into an abstract theoretical or conceptual issue, such as the one at stake in the Aristotelian friendship debate. Moreover, the debate has now branched out to include other kinds of platforms such as massively multiplayer online role-playing games (MMORPGs, see e.g. Munn 2012). My eventual aim is to refocus and extend the discourse even further by looking at sustained and deep email correspondence or epalship as a potential venue for the creation, development and maintenance of character friendships, and to do so by drawing an analogy with historically famous penpalships in the past. Most well-read people have heard about the unique bonds formed between penpals such as Harriet Taylor and John Stuart Mill, Johannes Brahms and Clara Schumann, and Voltaire and Catherine the Great. While I have yet to come across a paper explicitly arguing for those ‘bonds’ as ones of complete character friendship in the Aristotelian sense, it is difficult to shake the impression that the correspondence between those somewhat unlikely pairs of people comes as close as anything can get to exemplifying genuine and deep friendship of the most mature kind. In Section 4, I look closely at the friendship between Voltaire and Catherine the Great and explain the way in which it satisfies some of the salient criteria for Aristotelian character friendships. I then suggest, by analogy, that an epalship (email penpalship) between these two historic greats—had that been available to them at the time—would have offered at least an equally satisfactory outlet for character friendship.

A sceptic could claim that this argument by analogy constitutes at best a red herring and at worst a non sequitur, because it assumes that penpalships suffice as vehicles for character friendships, but that is something which representatives of the anti-camp mentioned above might flatly reject, with arguments similar to those mounted against modern-day online friendships. At the close of Section 4, I therefore argue that epalships allow for various technological extensions in the cyberworld of today that were not available to Voltaire and Catherine, for example, and that augmented with those extensions, there is even more reason for seeing epalships as potentially making the grade as true character friendships than traditional penpalships. Prior to the eventual aim noted above, a preliminary and subsidiary aim of this paper is to systematise in a new clarificatory way, and to offer a critical assessment of, the main arguments for and against online friendships as character friendships (in Section 3): an aim which is preceded (in Section 2) by a brisk tour of the relevant conceptual and empirical background, harking all the way back to Aristotle.

It is a sure sign of the enduring appeal of Aristotle’s highly moralised definition of the most supreme form of friendship that—with deferential historic stopovers in progenies such as Cicero and Emerson—the contemporary debate is still mostly couched in Aristotle’s terms and tied by various threads to both its merits and limitations. Some prominent aestheticised and amoral definitions notwithstanding (see e.g. Nehamas 2016), there is something intuitively attractive about the idea that the most mature form of friendship has to do with the mutual development and sustaining of virtue. However, real life rarely measures up to philosophers’ expectations, and I have argued elsewhere that Aristotelians, old and new, have been overly cavalier in accepting the claim that character friendships, once formed, have some sort of a unique enduring capacity that makes them immune to the wear and tear that often dissipates other forms of friendship (Kristjánsson 2019). So at the risk of losing some of the ground gained, I conclude by arguing, in Section 5, that despite being potentially categorisable as character friendships, mature epalships are vulnerable to the same problems and pitfalls as other examples of character friendships, and perhaps even more so. Despite that caveat, this paper is meant to make a positive contribution to the pro-camp regarding the question of the overall nature of online friendships as potential character friendships.

2 Some Problematics of Aristotelian Friendships in Online Contexts

Friendship for Aristotle is conscious ‘reciprocated goodwill’ (Aristotle 1985, p. 210 [1155b32–35]). It assumes three main types, where the first two (friendships for pleasure and utility) are ‘incomplete’—because of their essentially extrinsically valuable and transitory natures—but the most developed type (character friendships) is ‘complete’ because of its unique intrinsic and enduring capacity (1985, pp. 211ff [1156a6ff]). The two inferior types are not mere ersatz versions of character friendship, however. Pleasure and utility friendships have clear uses and are necessary for smooth human association, and they will continue even among those who have formed character friendships. While ‘base’ people can actualise the inferior types, but not the complete type, ‘good’ people enjoy all three types in different contexts but most specifically the complete type (1985, pp. 212 and 216 [1156b6–8 and 1157b1–4]). In all these three types of friendship, ‘you must wish [the friend] good for his own sake’ (1985, p. 210 [1155b31–32]), which distinguishes genuine friendship from mere friendliness, camaraderie or acquaintanceship. Later, Aristotle seems to contradict this claim by saying that those ‘who love each other for utility love the other not in himself, but in so far as they gain some good for themselves from him’ (1985, p. 211 [1156a11–12]), indicating that only character friendships make the grade as genuine friendships.

As it was not Aristotle’s wont to indulge in self-contradictions, exegetes have tried to save his face. I am in broad agreement with Cooper’s contention that Aristotle must here be focusing on different parts of the utility-friendship trajectory: its necessary instrumental formation versus its more noble development and maintenance. More precisely, Cooper takes Aristotle to be making the psychological claim that those who ‘have been mutually benefited through their common association will, as a result of the benefits […] tend to wish for and be willing to act in the interest of the other person’s good, independently of considerations of their own welfare’ (1977, pp. 633–634). Whiting (2006, p. 286) bolsters this interpretation with the general observation that ‘we cannot move immediately from the claim that a relationship comes to be for the sake of some end to the conclusion that the relationship continues to exist for the sake of that end’. I take it, then, that friendships for utility (and pleasure) are genuine types of friendship although only character friendship is ‘complete’ and intrinsically valuable. I say more later about why utility friendships tend to be undervalued in the relevant literature.

Apart from the features they share with the incomplete types, character friendships present some unique features of their own. For example, they involve (a) spending time together in shared activities, (b) sharing joys and sorrows, (c) loving the friend for her own sake in the special meaning of loving her moral character (as her set of virtues), (d) soulmateship in the strong sense of being ‘related to his friend as he is to himself, since the friend is another himself’ (1985, p. 246 [1166a30–33]; cf. pp. 260 and 265 [1170b6–7; 1172a32–34]) and (e) viewing the friendship as intrinsically valuable to the extent of seeing the friend as irreplaceable, not only painfully replaceable as in the more developed forms of utility friendships. Many of these unique conditions of character friendships are open to interpretation, however, as we witness, for example, for condition (a) in the following section. Moreover, in the case of (d), about the friend as a ‘second self’, it is not entirely clear whether Aristotle is speaking metaphorically or making a moral point about the essential substantive sharing of direction and purpose or making an ontological point about the inherently relational nature of selfhood. Condition, (c) of loving the friend’s moral character, seems to assume two forms or levels, of imperfect versus perfect ‘complete friendship’. In the case of budding phronomoi, on the one hand, what is at stake is the mutual love of imperfect ‘natural’ (habituated but not yet phronetic) developing virtue, which provides aspirations for further self-improvement on the way to full phronesis. In the case of fully perfected phronomoi, on the other hand, what we see is the mutual appreciation and affirmation of each other’s phronesis-guided-and-virtue-protected evaluative outlooks (Cooper 1977; Brewer 2005).

Against the context of this (perhaps) somewhat idealised description of the complete and perfect type of friendship, a lot of the current debate smacks of moral panic (see e.g. Deresiewicz 2009)—as if online friendships in general and those on Facebook in particular have plunged human friendship into a maelstrom of ephemerality, destroying its worth and turning it into a ‘toxic substance leading to isolation’ (Fröding and Peterson 2012, p. 201). Some of those hues and cries are directed at any genuine forms of friendship as reciprocated goodwill—the idea being (typically woven into a more general moral-declivity thesis about the cyberworld) that online friendships are replacing, corrupting or eroding the very constitution of friendships in the real world (Deresiewicz 2009). The debate relating to this broader question, which would overstretch the limits of the present paper, is subsequently about whether ‘the web is a boon for friendship, or a place where it falls apart’ (Vernon 2010, p. 12). In contrast, the more nuanced writings see online friendships as threats not so much to friendship per se as to Aristotelian character friendships in particular, with more and more people now allegedly settling for the inferior forms (McFall 2012)—meaning that online friendships, however pleasant or useful, may be damaging our ‘ability to develop as fully virtuous members of society’ (Sharp 2012, p. 231).

The advice from the robust naturalist Aristotle would definitely be to go out and inspect the empirical evidence for these claims. The problem is, however, that despite the rapidly growing mountain of empirical literature on the cyberworld and virtue (see e.g. overviews in Harrison 2014; Morgan et al. 2017), very little of it is geared towards online friendships in general, let alone Aristotelian online character friendships in particular (a concern raised, e.g. by Kaliarnta 2016). The latter lacuna is not unique to online friendships as putative character friendships. There is simply very little empirical evidence available about the three famous types of friendship, according to Aristotle, and how they do or do not play out in the real world (yet see Anderson and Fowers 2019). The study by Walker et al. (2016), which identifies unexpectedly frequent examples of presumed character friendships among children as young as 9–10, constitutes a notable exception, but unfortunately, this study does not address specifically the question of online friendships. We know that the frequency of self-reported friendships is on the rise among adults (Vernon 2010, p. 118). However, we must take such findings with a pinch of salt, both because the self-reports might be inauthentic as a result of our notorious lack of self-transparency, and because the frequency of friendships does not say anything about their quality and does not add argumentative weight, either way, to the Aristotelian debate about the relevant types of friendship and their respective worth.

Given the current state of friendship research, I am not sure we need to accept Vallor’s contention that ‘what philosophical and ethical discourse exists trails well behind the available empirical data’ (2012, p. 186). Firstly, the empirical data that would be relevant to the question of whether online friendships can constitute character friendships is simply not available. Secondly, even if such data existed or were gathered presently, it would in any case be accepted or rejected by scholars in the measure in which they accept or reject Aristotle’s conceptual taxonomy of friendship types. I would, therefore, argue that the lively debate about Aristotelian online friendships, to which I referred in the opening paragraph, has generally been carried out along sound methodological lines, although those have mostly been theoretical: by identifying features of Aristotle’s friendship types and arguing, typically by dint of intuitive or anecdotal evidence, about how those do or do not materialise in online friendships.

Despite being in broad agreement with the manner in which the current discourse has been conducted, I have identified four contestable assumptions that tend to be taken for granted by most friends and foes of online friendships qua character friendships: (1) (Aristotelian) virtue ethics is far better equipped to account for the moral salience of any deep friendships than ethical theories such as utilitarianism and deontology that demand impartiality. This is precisely why it matters so much for the defenders of online friendships to show that those can be brought within the rubric of Aristotle’s highest level with respect to virtue (see e.g. Vallor 2012, and Kaliarnta 2016, who otherwise disagree on whether this is possible). (2) The lower forms of friendship, especially friendship for utility, do not ‘merit the label “genuine friendship”’ and thus do not ‘qualify as morally valuable’ (Fröding and Peterson 2012, p. 201). (3) Only friendships between equals, and more specifically equals of ‘perfect’ virtue, count as true character friendships (McFall 2012, pp. 222–223; Bülow and Felix 2016, p. 24). (4) If we can demonstrate that (some) online friendships count as character friendships, we have demonstrated their true value as virtuous because such friendships are inherently stable and commendable from the perspective of wellbeing and virtuous living.

I consider those assumptions questionable either in themselves or as interpretations of Aristotle. Aristotle himself is the culprit behind the ‘inherently-stable’ part of assumption (4) because he seems to have been singularly oblivious to the exigencies that can threaten and dissolve even the most mature of character friendships (Kristjánsson 2019; I return to this issue in Section 5). He is not guilty for assumption (3), however, because two of the main threads running through his long discussion of character friendship are about how such friendship is also possible between unequals (given certain conditions), and how it plays a significant role in not-yet-perfectly-virtuous people helping each other make further progress on the road to full virtue—as already noted above. Aristotle did not have Mill and Kant to compete with so he remains understandably silent on assumption (1).

Assumption (2) is particularly damaging to the current debate, when couched in Aristotelian terms, about the moral value of online friendships. To recap, the underlying idea here is that for Aristotle ‘morally valuable’ means the same as ‘virtuous’ (see e.g. Fröding and Peterson 2012, p. 202). Because there was no word corresponding to the modern term ‘moral’ in ancient Greek, the impression may be falsely created that no developmental trajectories have moral value except those having to do with this ‘ethical’ (meaning in ancient Greek characterological) formation of stable traits (hexeis) of being, perceiving, feeling and acting. However, there is no doubt that Aristotle considered incontinence vastly superior to vice and continence to incontinence, although he had no term to term at his disposal to describe this superiority in modern terms as ‘moral’, relying rather on vague descriptors such as ‘characteristic of good people’ (see e.g. Aristotle 2007, p. 142 [1386b8–12]). I would argue that (some) types of Aristotelian utility friendships have indisputable moral value, on a modern understanding of ‘moral’, by helping friends turn vice into (at least) incontinence or incontinence into (at least) continence.

The upshot of the point I am making here is that Aristotelian friendships for utility are typically undervalued from a moral point of view in the friendship literature, and that if the aim is only to demonstrate that online friendships can have ‘moral value’, on a plausible modern terminological reordering of Aristotle’s developmental account, then the bar is set too high by focusing solely on character friendships. That said, the present paper aims higher than simply noting this point. I argue that some forms of online friendships have the potential to satisfy the stringent demands of character friendships, and I demonstrate that with the example of epalships as augmented penpalships in Section 4. Prior to that, however, it is instructive to rehearse briefly the standard arguments against the overall possibility of Aristotelian online character friendships.

3 For and Against Online Aristotelian Character Friendships

Given the quantity and quality of the general discourse on arguments for and against the possibility of online ‘character friendships’ in the Aristotelian sense, I only offer a brief systematisation and critical classification in the present section, with the odd additional observation, and I limit myself mostly to consideration that I can revisit serviceably in the following section where I argue the case for penpalships/epalships.

Although I do consider the anti-arguments largely misplaced, they serve as a useful springboard for the discussion and provide considerable food for thought. I divide them into geographical/ontological, epistemological, psychological and social arguments, respectively, and explore each in turn.

The Geographical/Ontological Argument(s)

This argument is also known as the ‘physicality objection’ (see e.g. Elder 2014). There are two ways to understand it: as a complaint about online friendships lacking the geographical proximity of two human persons in time and space, or as a deeper philosophical worry about the ontological distance between online friends, preventing the formation of friends as ‘second selves’ on Aristotle’s (notably fairly ambiguous) understanding. This is why I use a double-barrelled designator for the argument in question. The valid empirical point that most online friendships still just complement real-life ones (see e.g. Vallor 2012, p. 186) does not blunt the force of this argument. If it is true that online friendships fail to make the grade as character friendships without such physical complementarity, that condition rules out of court examples of pure penpalships/epalships like the ones I draw upon in the following section. The geographical/ontological argument (esp. when eliciting the idea of ‘multi-filtered communication’, see below) often shades subtly into an epistemological argument, but I will try to prise the two apart in what follows.

The premise of the geographical argument is unambiguous. It relies on Aristotle’s strict condition that character friends must ‘live together’ and ‘share conversation and thought’ (1985, p. 261 [1170b10–15]). The standard interpretation is to see this as requiring actual physical cohabitation and mutual participation in shared bodily activities. However, a couple of rejoinders have been made, or can be made. (a) The standard ploy is to invoke Aristotle’s own caveat that living together does not mean sharing the same pasture like grazing animals (1985, p. 261 [1170b13–15]; see e.g. Elder 2014). Had Aristotle been aware of contemporary possibilities of sharing lives online, he might precisely have taken those as examples of mutual participation in activities that do not require ‘grazing’ in the same field. As Elder (2014) observes, sharing a conversation about one’s day with a friend online could well count as ‘living together’. While Elder concedes that people might be badly off, friends-wise, if none of their friendships involved physical presence, this does not mean that particular friendships with merely online presence cannot count as (character) friendships. (b) While examples of shared activities that Aristotle takes include gymnastics, hunting and playing dice, the fundamental activity that Aristotle considers character building, and hence constitutive of character friendships, is engaging in philosophical conversation together. As Pangle notes, this is not so much because philosophy is ‘grave and the others light-hearted, as because it is a deeper, more satisfying, more unmixed, and more lasting’ as pleasure-and-flow-yielding activity (Pangle 2003, p. 195). Arguably, sharing thoughts stemming from contemplative activity (theoria) is something to which online media lend themselves well. Fröding and Peterson would object that friends spending time together in real life are ‘more likely to face a wider spectrum of different situations, and consequently, encounter a wider range of topics meriting contemplation’ (Fröding and Peterson 2012, p. 204). However, this is a dubious empirical claim. I would contend that I encounter more thought-provoking materials on Facebook, in all the exciting articles and news items posted by my philosophical friends, than I would be likely to come across playing dice with them—or even going hunting.

The more sophisticated rendering of this specific anti-argument is to focus on ontological rather than mere physical distance. McFall (2012, p. 224) offers an incisive consideration of this sort with his distinction between multi-filtered communication (where information about A’s experiences is first filtered through A’s [online, text-based] interpretation before being reflected to B), and single-filtered communication (where B has direct and simultaneous real-life access to A’s experiences). McFall’s assumption is that in single-filtered communication, ‘nothing stands between A and B’ and they can ‘directly perceive each other’ (McFall 2012, p. 224). One could read this assumption epistemologically, about the limits that online media place on knowledge transfer and how they can facilitate knowledge manipulation; I return to those concerns below. However, it is more instructive here to understand the assumption ontologically, about multi-filtered communication creating an ontological barrier between two people which makes it impossible for them to penetrate each other’s selfhoods in the way that Aristotle’s ‘second-self’ condition on character friendships requires. But interpreted in this way, the assumption reflects a strong positive bias towards the body (cf. Bülow and Felix 2016, p. 31). As the German saying has it, ‘loneliness alongside another person is the worst kind of loneliness’. Unfortunately, there are no immaculate perceptions of other people’s selves available to us, no inherently un-warped mirrors that enable us to merge with them in a strong form of soulmateship, simply by being placed next to them in time and space. If there is any such thing as the ontological sharing of selfhoods, it is not achieved through some sort of direct physical access to the innermost core of another person but rather through finding ways to navigate through the various filters that inevitably stand between any two persons, in order to co-create an extended self, or a self that is enlarged through emotional and intellectual attachments. We are constantly interpreting our own thoughts and actions to ourselves and to our friends, even when we are engaging in the most physically intimate acts of friendships with them, and who we think we are and how we self-interpret is partly constitutive of who we really are (Kristjánsson 2010). Ontological extensions of selfhood—if there is any such thing—would surely be dependent upon many necessary variables, but actual physical cohabitation is, arguably, not one of them.

The Epistemological Argument

This argument (explored e.g. by Fröding and Peterson 2012 Kaliarnta 2016, and Bülow and Felix 2016) can most helpfully be understood as being two-pronged. The first prong has to do with the limited information that online communication typically conveys about who we are, and a selection bias regarding the information that we typically choose to make available to the online receiver(s)—not because we deliberately aim to deceive them but because we tend to disseminate this information with a particular purpose in mind and do not receive the ordinary prompts from face-to-face contact inducing us to explain and expand upon the information provided (see e.g. Sharp 2012, on ‘unintentional obfuscation’, p. 233). There is not much that needs to be said about this argument here except that it is usually mounted in the context of discussions about social-media (Facebook-type) friendships. However, to the best of my knowledge, no one has ever seriously argued that typical Facebook friendships constitute character friendships (although they may possibly complement them). Deeper and more multi-faceted online friendships, as I explore in the following section, seem immune to this first prong of the epistemological argument; hence, it does not cast any serious chill over the idea of online character friendships.

A more explicit and in-your-face version of this argument is to frame it as a ‘deceptiveness objection’ (as elaborated upon and responded to by Elder 2014). Online media allow people to engage in impression management and deliberately distorted self-representations that they would never get away with in the real world because people—at least their close friends—would see through such misrepresentations. There may be various psychological mechanisms behind the intention to deceive—ranging from dissociative imagination to a more sinister moral disinhibition effect (Suler 2004)—but the upshot is the same: online self-representations are too easily amenable to manipulation to allow for the epistemic transparency that needs to be at issue between true character friends.

One response to this argument is to bite the bullet and acknowledge that it is, in principle, easier to engage in distorted and misleading identity construction online than face to face. However, perhaps there is less of a need to do so. The ‘games’ we play with people face to face, in projecting who we are, are often motivated by shyness and insecurity. The leisurely pace of some online communications, such as emails, may allow for more authenticity, candidness and deliberateness, as we are not put on the spot to reveal ourselves to the potential friend in a panicky way, without the necessary prior deliberation about how to express ourselves in the most appropriate manner (cf. Briggle 2008). Another and quite different response is simply to reject empirically the claim that deceptions are easier to administer in online (text-based) communications. The assumption behind the epistemological argument may again reveal a positive bias towards the body as more revelatory (less ‘filtered’, recall above) of who we are than what we write. However, sociopaths and con artists are known to be particularly apt at using facial cues and bodily charm to mislead. In contrast, text correspondence leaves digital paper trails that can be revisited and ‘deciphered’ in an orderly and un-hasty fashion (Elder 2014). In any case, the epistemological argument—when couched in terms of the potential for deliberate deceptions—while possibly hitting at the formation of character friendships in the earliest stages, surely leaves mature character friendships intact, because true character friends would have no incentives for trying to cloud one another’s vision of who they are ‘deep down’.

The Psychological Argument

Recall that one of Aristotle’s conditions for character friendships is that the friends—as soulmates—share joys and sorrows. However, a common complaint about online communication is that it waters down and degrades the expressions of ‘being happy for’ or ‘empathising with’ that human beings have learned to do so well, through a long evolutionary process, in face-to-face encounters. While empirical evidence shows that empathy is often expressed online (see e.g. Morgan et al. 2017), the worry would be that there is something about such expressions that makes them superficial, shallow and lacking in the intimacy that comes to the fore when we actually embrace a friend in the flesh. In that sense, online friendships would be inferior with respect to some of the virtues that ex hypothesi make character friendships ‘complete’ (by Aristotle’s lights) and unique (see e.g. Vallor 2012).

As Elder (2014) correctly notes, the psychological argument (which she refers to as the ‘superficiality objection’) is most often mounted against friendships forged on social-media platforms, and it may leave other online-communication alternatives untouched. Rather than expanding on that point—although Section 4 below obliquely does so—I propose to take a different tack here and focus on a specific feature of Aristotle’s theory of virtue and virtue education that has rarely been elicited in the discourse on online character friendships. I am referring here to Aristotle’s insistence on the individualisation of virtue and education for good character (be it formal education or education through encounters with character friends). Different people have different needs and tendencies, and hence the proper (medial) state of character ‘is not the same for everyone’ and ‘in the object’, but rather ‘relative to us’. For instance, what is moderation in eating for me is not the same for Milo the athlete (1985, pp. 42–43 [1106a30–b5]). Similarly, ‘each state [of character] has its own special [view of] what is fine and pleasant’ (1985, p. 65 [1113b31– 33]), and a boxing instructor will not ‘impose the same way of fighting on everyone’ (1985, p. 295 [1180b9–11]).

The upshot of these comments is that what counts as the proper sharing of joys and sorrows will depend essentially upon the individual constitution of the friends in question and how they exemplify the relevant virtues in their own personalised way. Contra McFall (2012, p. 222), there is arguably nothing in the nature of online means per se that diminishes the opportunity for robust moral reflection and the virtuous improvement of self and the friend qua ‘second self’. To rely on a personal anecdote, I happen to be a person to whom acts of physical intimacy (embraces, hugs, high-fives) with close friends do not come easily. I even find myself inhibited from showing many of my deepest emotions to them face to face. However, I have engaged in lengthy email correspondence with a number of friends, including ones who have gone through serious trials and tribulations, and I am very proud of the feedback I have got back from some of them, talking about the emotionally deep reflections in my emails as ‘life-saving’ and ‘epiphanic’. Of course, some of this feedback might simply be polite gratitude talk. Be that as it may, I do identify with the character ‘Pete’ in Bülow and Felix’s paper who referred to the online world as ‘where I feel most myself’ (Bülow and Felix 2016, p. 27).

As a moral ‘boxer’—to paraphrase Aristotle—my ideal ‘way of fighting’ is to exhibit my virtues via emails and texts. The psychological argument claims that there is something superficial and shallow about those means of communication. Dissenting, I would argue that I have found my virtue niche in exploiting those technological vehicles as conduits for who I am deep down, and how I convey to my closest friends the lessons that express my ‘wishing them good for their own sake’, as Aristotelian character friendship requires.

The Social Argument

According to this argument, there is something socially faulty about the platforms on which online friendships take place. What is considered ‘faulty’ is usually the commercial nature of those platforms and how they represent ulterior motives for trying to sell us products (or even pitching certain political causes) at the same time as creating spaces for friendships; this is why Elder (2014) refers to this argument as the ‘commercialization objection’. This concern is the one least potentially detrimental to the positive argument that I propose to make in the rest of the paper. It suffices, therefore, to note here that the standard venues where deep friendships have historically been pursued tend to be commercial in nature; consider tennis clubs, sport stadiums, pubs and restaurants. It seems implausible to think that just because someone has commercial interests in these venues, this undermines the quality of friendships pursued within them (although one might worry if all friendships were developed in environments of a commercial or profit-seeking nature).

To conclude this section, someone could argue that the whole discourse on the deepest and most profound friendships is cluttered with baggage from Aristotle’s idiosyncratic concept of ‘character friendships’ and that we might be better off dispensing with that concept altogether. While I go on to acknowledge some chinks in the armour of this concept in Section 5, the present section has failed to demonstrate that the standard general arguments against the possibility of Aristotelian character friendships being pursued online do work. It is now time to turn from those ’negative’ arguments to the ‘positive’ one about one form of online friendship that I do think can pass muster as true character friendship: a sophisticated and expanded form of epalship.

4 Penpalship as Character Friendship and How It Can Be Augmented as State-of-the-Art Epalship

Penpalship is a historically established form of communication and friendship which has now been mostly overtaken by email correspondence and other forms of epalship. It is a matter for some surprise and disappointment that neither old-fashioned penpalships nor contemporary epalships seem to have been subjected to sustained academic scrutiny as potential exemplifications of (Aristotelian) character friendship in the same way as general online friendships, especially on social-media platforms, have been in recent times. I happened to be an active participant in penpalships before the days of email, in particular with friends I had met at university and to whom the adage ‘out of sight, out of mind’ did not apply. More recently, I have engaged in extended epalships with a small group of people, including one person P whom I have only fleetingly met twice, as P happens to suffer from extreme anxiety, caused by a borderline personality disorder, which makes ordinary face-to-face interactions extremely taxing for P. P happens to be highly intelligent, articulate and morally reflective, and I would argue that our correspondence has had a significant morally edifying influence on both of us. Whether the moral reform in question is at the level of character friendship, or of the sort of continence-enhancing utility friendship that I mentioned as also being morally valuable in Section 2 is open to debate. In any case, it is more instructive here to rely on a well historically documented form of penpalship that one would be loath to describe as anything but full-blown character friendship: that between Voltaire and Catherine the Great.

To be fair, it is not as if penpalship or epalship are never mentioned in the literatures on character friendship in general and (putative) online character friendship in particular. Barnard (2011, pp. 167–168) observes how letters between friends allow for the kind of ‘fine-tuning, for carving out a space for your personality’ that spontaneous face-to-face interactions rarely do. In addition to the possibility of reflective depth, many writers mention how email platforms become a helpful levelling field for people who suffer from physical or psychological impairments of various sorts—or simply shyness and social withdrawal—that prevent them from cultivating real-life friendships (see e.g. Kaliarnta 2016; cf. also Schaubert’s 2019 moving example of the deep online friendships of a severely disabled boy to which I return later). However, the received wisdom seems to be, as suggested by Elder (2014), that because epalships are so remarkably like penpalships, they may not offer much that is distinctive enough for theorists engaging with emerging technologies to think about. I consider this suggestion misleading in two ways, both because penpalships themselves have not been scrutinised systematically, to the best of my knowledge, as forms of Aristotelian character friendships, and because epalships offer possibilities nowadays that go far above and beyond those of mere penpalships.

The great Enlightenment philosopher Voltaire (1694–1778) was a prolific letter writer, leaving behind him at least 20,000 pieces of letter writing. His most well-known and sustained correspondence was with Catherine II (1729–1796), commonly referred to as ‘Catherine the Great’: empress of Russia 1762–1796. According to the close historical study by Reece Stuart (1914), on which most of the facts about this penpalship elicited here are drawn (apart from obvious web sources such as Wikipedia), their correspondence began in 1763 with a letter from Catherine. The collection of letters includes 74 from Catherine to Voltaire and 91 the other way round. The liveliest period of their penpalship was between 1770 and 1771, when all in all 71 letters were sent. The most conspicuous fact about this extensive correspondence is that the pair never met in real life. This marks their penpalship out from other famous historical examples, such as that between Harriet Taylor and John Stuart Mill, which also included face-to-face interactions and ended with their hooking up (although the eventual formal relationship continued, famously, to be more intellectually than romantically stimulated).

In the letters between Voltaire and Catherine all imaginable subjects of the day are touched upon, from philosophy, art, literature and politics to more practical and mundane matters. No visible signs of self-censorship or deliberate lack of self-transparency can be noted. The pair are incredibly forthright about all matters that arrest their attention, and although many of the issues they discuss are both politically and personally sensitive, they seem to speak about them as if the possibility of these letters ending up in the wrong hands had never occurred to them. If there ever were true soulmates, then Voltaire and Catherine seem to satisfy that description admirably. However, to establish that their soulmateship was genuinely a case of Aristotelian character friendship, it is helpful to revisit some of the criteria of such friendship from Section 2.

Shared Activities

While establishing the precise conditions for an activity to be ‘shared’ in an Aristotelian sense is a tall order, as we saw above, there is no denying the fact that Voltaire and Catherine engaged in shared philosophical discussions about common topics. They also almost constantly exchanged gifts and favours, in a very balanced way. The ‘favours’ included ‘books, translations, codes, nuts and plants from Siberia, advice, paintings, portraits, watches and other jewellery’ as well as ‘employment of friends’ (Stuart 1914, p. 66). Finally, they encouraged and closely followed each other’s activities in various literary, political and humanitarian endeavours. To be sure, they never went hunting together or played dice, but given Aristotle’s preoccupation with the shared activities of theoria, it is difficult to see why the activities of the pair in question should fail to count as ‘shared’.

Sharing Joys and Sorrows

It may be difficult to realise this at first glance, because of the elaborate layers of feigned politeness and flattery that characterised all letter writing between learned people in the 18th century, but as Stuart (1914, p. 69) notes, when peeling these layers away, it becomes obvious how a ‘very natural bond of sympathy’ gradually developed and became sealed between these penpals. The utterances of mutual care and love are astoundingly personal, given codes of courtesy in letter writing of the day: for example, Catherine’s genuine concerns about Voltaire’s failing health and his concerns about her difficulties in putting some of her more ambitious philanthropic projects into action. Her eventual mourning for Voltaire’s death comes straight from the heart: ‘a great man who loved me so much’ (Stuart 1914, p. 87).

Loving the Other qua That Person’s Character

This is perhaps the most salient criterion of character friendship, as it determines precisely whether the two friends are mature enough, and brought up well enough characterologically, to count as either budding phronimoi or full-fledged phronimoi, capable of loving and learning from each other’s virtues. Now, without being able to penetrate into the hearts of either Voltaire or Catherine, it would seem churlish in the extreme to deny them the status of at least developing phronimoi. The tenor of the discussion in their letters is not about how to help each other to force themselves, against the thrust of incontinence, to become good through self-control or continence, let alone to avoid vice, but rather how to make themselves more virtuous and the world a better place. It could even be argued that if any famous 18th-century historical figures deserve to be placed at the level above ordinary phronimoi in Aristotle’s system, namely the level of megalopsychoi, Voltaire and Catherine should occupy those places. The megalopsychoi—or great-hearted persons—possess all the same virtues as the phronimoi, but at an advanced level of attainment and exhibition because of their access to resources that enable them to become public benefactors and engage in great deeds of philanthropy. While not an unmixed blessing, to be a megalopsychos gives a person unusual powers of moral reform and public grandeur (see Kristjánsson 2020, chap. 4).

It is clear that both Voltaire and Catherine saw their roles as public benefactors, witness for example the empress’s commitment to modern medicine and public health by allowing herself to be inoculated against smallpox as a ‘guinea pig’. She also stablished a schoolhome for nearly 40,000 children who had been abandoned by their parents. Her eagerness for acting as a social entrepreneur is beyond doubt, although her practical endeavours were not always successful, especially during the latter part of her reign. She was also considered ‘mannish’ or ‘masculine’ in a sense similar to that which Aristotle seems to ascribe to his megalopsychoi with their deep voices and alpha-male comportment. Voltaire’s main contribution to the advancement of humanity may have been intellectual, but he was also keenly interested in more practical aspirations, as witnessed by the watch factory he built at Ferney to give employment to refugees from Geneva. Not the greatest of salespersons, he had to rely on his friends, most notably Catherine herself, to buy most of the watches. In particular during the last years of his life, Voltaire became increasingly concerned with the welfare of the region around him. All these activities by our two protagonists mutually inspired each one of them in ways that cannot be better described than having been conducive to the cultivation of virtue—and they loved each other for all these gestures and activities as representative of who they were deep down.

The Character Friend as a Second Self

At least during the most intensive period of their letter-writing and soulmateship, Voltaire and Catherine seem to have seen their lives as inseparably intertwined, despite the geographical distance. Their individual projects, such as Voltaire’s infamous watch factory, became ‘their’ joint projects. Their friendship was dynamic and reactive, and no feat was considered complete until it had been talked through with the other person. This penpalship demonstrates the sort of joint affirmation of moral and aesthetic sensibilities that Aristotle seems to have been referring to with his ‘second-self’ thesis (cf. Brewer 2005). If the notion of ‘extended selfhood’ has any meaning at all, beyond the metaphorical, then it seems to have been instantiated in this uniquely deep penpalship.

The Friend as Intrinsically Valuable and Irreplaceable

Voltaire had such a high opinion of Catherine that he called her several times ‘empress Saint Catherine’. Similarly, Catherine referred to Voltaire with undivided attention at ‘the man to whom I owe all that I know and all that I am’ (Stuart 1914, p. 88). She had in her library a bust of Voltaire. Conversely, he possessed a portrait of her which he called a shrine. Two related caveats may be entered here, however, about this condition of intrinsic value and irreplaceability. One is that a moral mentor/educator/exemplar may also count as irreplaceable in Aristotle’s system, not only a character friend, and there is no denying the fact that, in the first instance, Catherine approached Voltaire very much as her superior. She admired him as a mentor rather than loved him as an intrinsically valuable character friend. However, those dynamics seem to have changed fairly quickly, and once the relationship between them became more on an equal footing, their value to one another became much more as that of an irreplaceable friend than an irreplaceable mentor and mentee.

Secondly, Aristotle is very sensitive to the complex power relations that exist between character friends. Notably, despite their putative character friendship, Catherine obviously remained more powerful than Voltaire politically, on the world stage, and he had more symbolic capital as an intellectual figure. For those, who like McFall (2012) and Bülow and Felix (2016) think that true character friendship can only obtain between equals (recall the assumption critiqued in Section 2), this might undermine the specific example of penpalship as character friendship that I have been exploiting in this section. However, Aristotle does allow for true character friendships between unequals, both in terms of social status and virtue development. To explain those, he offers us his principle of proportionality, or ‘equalising principle’, according to which, in unequal friendships, the stronger party ‘must be loved more than he loves; for when the loving reflects the comparative worth of friends, equality is achieved in a way’ (1985, p. 221 [1158b26–29]). So while the weaker party gets ‘more profit’ (presumably in terms of character growth), the stronger party gets more honour and devotion (1985, pp. 236–237 [1163b1–15]). It is thus ‘proportion that equalizes and preserves the [unequal] friendship’ (1985, p. 238 [1163b34]).

I am not sure that we need this principle here in terms of social status, because the superiority of Voltaire on the intellectual stage and Catherine on the political stage may well have levelled each other out. However, even if we argued that Voltaire remained superior to Catherine in his attainment and display of reflective virtue development (which is the sort of inequality that seems to worry McFall 2012, and Bülow and Felix 2016, rather than the social status one), we could apply Aristotle’s equalising principle to show that, as long as Catherine gained more morally from her correspondence with Voltaire, and she showed him more honour and deference in the letters, which actually seems to have been the case, their penpalship can still count as one of true character friendship, albeit unequal in a certain way (cf. Cooper 1977, on true character friendships between people unequal in virtue development). The gender issue (that theirs was a cross-sex friendship) complicates this conclusion somewhat, however, from an Aristotelian perspective, as I will explain in the final section.

Do any of the general misgivings about character friendships between people who do not meet in real life (rehearsed in Section 3 above) threaten to undermine this conclusion? The social argument is not relevant here, but let me offer brief comments about the other three. According to the geographical/ontological argument(s), distance in space creates barriers to full self-disclosure and the merging of selves. It is very difficult to detect any such barriers in the penpalship between Voltaire and Catherine. Indeed, a comment about another penpalship would seem to apply here: through ‘absence, geographical distance, and the written word’, yet ‘somehow these two people created out of words a nearness we today do not entirely grasp’ (Vernon 2010, p. 116). The epistemological argument focuses on the limiting or intentionally deceiving nature of the information channelled between friends who do not meet in real life. The unerring mark of the correspondence between Voltaire and Catherine—despite the socially expected flattery—is, however, their ‘sincerity’ and apparent lack of ‘some ulterior motive’ (Stuart 1914, p. 86). The necessary bluntness of correcting each other’s perceived misconceptions is also there, although sometimes well concealed as ‘a drop of acid in a bonbon’ (Stuart 1914, p. 68). What is so charming about the whole collection of these letters is precisely their unmitigated authenticity. There are no half-hearted evasions, no half-measures. Finally, I would challenge anyone to identify an aura of shallowness and superficiality in this correspondence, as suggested by the psychological argument. The trajectory of the letters shows increasing levels of personal intimacy (Stuart 1914, p. 69). There is no fickleness or inconstancy, and the occasional gaiety of phrasings is representative of the deep spiritual bonds that existed between this unlikely pair of people, even including a shared sense of humour.

All this extended and multi-faceted argument about the penpalship between Voltaire and Catherine as (Aristotelian) character friendship notwithstanding, I can still envisage misgivings to the effect that although some measure of character friendship had been reached in this penpalship (and others of a similar ilk), there was still something missing to perfect it and make it fully ‘complete’ in the Aristotelian sense. Would it not have been even better had they been able to meet and communicate more directly?

Those misgivings give me the opportunity to return more explicitly to the main theme of this paper, about contemporary online friendships as potential character friendships, by suggesting ways in which modern technologies would have made the penpalship between Voltaire and Catherine even more edifying and gratifying. Obviously, their correspondence was hampered by a slow postal service and the impossibility of immediate feedback. However, modern-day epalships allow for a number of extensions that would augment historic penpalships: rendering it even more plausible to consider them passing as true character friendships.

  • Contra Barnard (2011, p. 164), email correspondence can function exactly like old-fashioned penpalship if the parties so wish. There is thus nothing in the nature of email correspondence per se that automatically causes cognitive shallowing by foreclosing the option of a leisurely pace of corresponding, deep reflection, forethought and finessing before the correspondence is sent. However, emails are received earlier than letters and can also be responded to immediately if that is deemed necessary. Given the immediacy of many of the projects in which Voltaire and Catherine were engaged, I am sure they would have loved the option of spontaneous responses and also of reverting to messaging platforms like Whatsapp for shorter and quicker forms of communication. They would most certainly have become Facebook friends and joined special interest groups on Facebook.

  • For those who still believe facial cues are vital to decipher the mind of the friend qua ‘second self’, video-based platforms like Skype are an ideal substitute for real-life meets. I conjecture that Voltaire and Catherine would have Skyped each other quite often.

  • I am also sure that both Voltaire and Catherine would have relished the chance to blog and give TED talks, and then to correspond or Skype about those.

  • Both Voltaire and Catherine were avid lovers of art. Modern technologies allow people to visit virtual museums on the web and even do so simultaneously, exchanging views in real time on what they see and experience. This would have been a heaven-sent option for our two famous penpals.

  • Multiplayer games or, more specifically, MMORPGs (where people interact through avatars in virtual worlds) provide means of ‘shared activities’ in the literal sense. Many of those games test the participants’ moral fibre in various ways and help them identify automatic emotional responses that they may not even be aware of themselves. Such games thus afford various opportunities by which to enhance both self-knowledge and other-knowledge. There is a cogent argument in Munn’s 2012 paper for why MMORPGs satisfy Aristotle’s criteria for shared activities, goods and interests more so than other online platforms. Munn argues that such games obliterate a clear distinction between the physical and virtual worlds and that they facilitate stronger social bonds between participants than most, if not all, other online activities. For a particularly striking and moving illustration of putative character friendships pursued in MMORPGs, I recommend Schaubert’s 2019 account of Duchenne- muscular-dystrophy sufferer Mats Steen (aka Lord Ibelin Redmoore) who spent up to 20,000 hours (equivalent to more than 10 years’ full-time employment) in World of Warcraft (WoW) and made many deep and close friends there, before his untimely death.

Given Voltaire’s and Catherine’s common interest in the development of virtuous character, they would arguably have benefited from, and made use of, multiplayer games to get to know each other even better and to help each other grow in virtue. There is another reason why I have kept this point as the last, and perhaps best, example of how Voltaire and Cartherine could have augmented their penpalship if they were living today. Many MMORPGs, such as WoW, involve warfare, and both Catherine and Voltaire were deeply interested in questions of war and peace and how wars should ideally be justified and conducted. WoW would have been an ideal venue for them to cultivate those shared interests in an interactive way.

These are just a few scattered examples which show how contemporary forms of epalship can be extended to turn the psycho-moral mechanisms behind old-fashioned penpalships into even more productive vehicles for character friendships. Should any of the misgivings from Section 3 still remain with respect to mere penpalships, I think they will melt away once we consider the possible modern augmentations. Moreover, and as importantly, all these new technological options would have resonated well with the lively and curiosity-driven Enlightenment spirit of Voltaire and Catherine the Great.

5 Concluding Remarks

The early onslaught on the idea of online friendships as (Aristotelian) character friendships provoked an eruption of helpful responses, as already noted. My sympathies with those responses notwithstanding, I hope to have added to the argumentative arsenal by showing how the fairly simple and time-honoured form of friendship called penpalship could, even before the days of technological wonders, pass muster as character friendship, and how that categorisation is even more apt today as penpalships have shaded into contemporary epalships. I would thus conclude, to cite Barnard (2011, p. 118), that ‘contrary to all the nay-sayers and doom-mongers, technological developments are injecting a new dynamism into [character] friendship’. Some of the sceptical writings about online friendships as character friendships smack of a nostalgic elegy on the disappearance of a world in which human relations were deeper and character friendships easier to establish. The example of Voltaire and Catherine shows, however, in my view, that their character friendship thrived in spite of, but not because of, the low-tech means of communication that were available to them. Modern technology would have made it so much easier to cultivate, maintain and deepen.

All that said, I do not want to be seen to be replacing an elegy with a panegyric on how emancipatory technological innovations have taken character friendships to a new higher qualitative level. I have not been arguing that online friendships have some unique qualities that make them stand above the fray. For example, for every successful penpalship/epalship that migrates easily into real-life encounters, there are probably just as many that break down once the two friends actually meet, witness well-known historical examples of Groucho Marx and T. S. Eliot, on the one hand, and Rousseau and David Hume, on the other. I have simply been making the more modest claim that epalships can satisfy the conditions of Aristotelian character friendship and that standard misgivings about online friendships as character friendships do not necessarily hit at them.

In a previous paper (Kristjánsson 2019), I lamented how—in contrast to the rest of the Nicomachean Ethics where Aristotle is keenly aware of venturing over the terrain of an inexact science, mined with various counter-examples (1985, p. 36 1104a10–11])—his exploration of character friendships is strangely devoid of examples of cases that may debar, stunt, decimate and ultimately dissolve such friendships, and how to tackle those. Admittedly, there are long sections addressing problems and difficulties in friendships more generally (see esp. 1985, pp. 232–243 [1162b1–1165a35]), but those are almost entirely focused on the two lower kinds of friendship (for pleasure and utility), suggesting that once elevated to the higher level of character friendship, these problems will vanish. Aristotle did famously acknowledge that character friendship could be terminated if one of the parties turns ‘incurably vicious’ (1985, p. 244 [1165b15–35]). However, in the previous paper, I listed various other potential pitfalls and problems of character friendships. I will only briefly mention two here that I think might be even more difficult to deal with or remedy in even the most advanced forms of epalships than in ordinary real-life character friendships.

The first is possible eros–philia conflict. Aristotle seems to be completely insensitive to the faint undercurrents of excitement often floating close to the surface in friendships between those character friends who could potentially be sexually attracted to one another. It is difficult to untangle the extent to which the friendship between Voltaire and Catherine may have had romantic undertones. At the time their correspondence began, Voltaire was already a relatively old man (although never too old to fall in love!). It is sometimes hard to distinguish between flattery and flirtations in their correspondence (Stuart 1914, p. 70). We do not know whether one or both of the parties were romantically motivated. However, eros can be potentially detrimental to character friendships and create ineliminable tensions, especially when one of the parties is not keen on anything beyond philia. Mutual uncertainty can be particularly damaging here. Human beings have evolved such as to be able to pick up signs of sexual interest in the other person’s bodily movements and facial expressions—features which may not reveal themselves as well even in video calls as they would in face-to-face encounters.

It could be argued that the reason Aristotle does not address eros–philia conflicts is that his discussion of philia is exclusively about friendships between male (budding or fully fledged) phronimoi. This would then also make my long-drawn-out example of Voltaire and Catherine the Great rather ill-suited to the purpose of illustrating specifically (online) Aristotelian character friendship. Although Aristotle leaves open the possibility of character friendships between husbands and wives, ‘if they are decent’ (1985, p. 232 [1162a25–26]), it is quite true that the intended readership of the Nicomachean Ethics comprises well brought-up men. However, eros–philia conflicts between older and younger men was a well-known theme in Aristotle’s time (witness e.g. Plato’s Symposium), so this does not explain Aristotle’s conspicuous silence on the matter. I chose the example of the penpalship between Voltaire and Catherine because of how well-known and well-documented it is, although it does invite complications as an ‘Aristotelian’ example. There is a more ‘Aristotelian sounding’ example of penpalship, which appears to pass muster as character friendship, in Briggle’s paper (2012, pp. 73–74), about the friendship between a soldier in the Civil War and a male Boston school teacher, whom he has never met—but the problem with that example is that it is fictitious and hence easier to ‘manipulate’ to serve the purpose of the argument.

The second problem I want to address in this final section is that of our strong need for self-verification. People tend to be in the business of ‘seeking to confirm their own view of themselves’ (Aristotle 1985, p. 222 [1159a21–24]). Since one of the fundamental roles of character friendships is the mutual correction of evaluative outlooks and virtuous make-up, in order to help one another on the way to phronetic virtue, this may require considerable doses of nudging and even admonition. However, people—even true character friends—may not always take well to having their view of themselves challenged. In face-to-face encounters, human beings have evolved intricate strategies to do this in measured ways, through constant feedback from the other party, in order to gauge whether the criticism is too ‘soft’ or too ‘hard’. This will be more difficult to decipher, in a measured way, in text-based communication, even when complemented with the odd video call.

So there may, after all, be specific problems that stand in the way of epalships instantiating true character friendships. However, the two issues mentioned here are just intensified versions of problems that already exist in ordinary real-life character friendships, and they do not impugn the overall diagnosis offered in this paper of advanced epalships as potential forms of character friendships.