Keeping company: Educating for online friendship

The possibility of online Aristotelian virtue friendships via social network sites continues to be raised by philosophers, but as yet this has not been positioned within the realm of children or adolescents, who are known to be amongst the largest users of social media Governmental agencies across the globe still struggle to define the boundaries of online usage for children, often depending on school-based curricula highlighting ?safe-guarding? online or some form of character education This, however, often leaves the philosophical thinking behind virtual relationships as incompletely addressed in educational theory, policy and practice, despite there being some very real difficulties for children Utilising the insights of Aristotle on friendship, I offer a view that may hold potential for a philosophically based policy I outline three different ways this philosophical literature could have implications for education and indicate the types of policy that each might entail I will contend that there are three distinct stakeholders here that can be identified as having a significant role to play in what we should do: the schools themselves, educational researchers and policy writers Finally, I suggest ways in which research, policy and practice might link together


Introduction
Discussions, both scholarly and amongst the general public through the news media, around how schools should approach the fundamental aims and overall principles of educating for social media usage have become increasingly complex in recent years (e.g. Trujillo-Jenks & Jenks, 2015;Krutka & Carpenter, 2016;Milosevic & Livingstone, 2017). This has been complicated by an array of findings that continually demand vigilance against 'the dark side' of the virtual world, particularly for those considered vulnerable in some way: the prevalence of internet predators and the grooming of children; the rise in 'trolling' and other online bullying behaviours; increased levels of child mental health issues being related to social media usage (Livingstone, 2014;Sadler et al., 2018, amongst others). A lack of dependable research has made it difficult to systematically address the impact of these problems with children: whilst most agree that something should be done, what that should be, and by whom, is often still debated. Indeed, governmental agencies across the globe still struggle to define, and thus patrol, the boundaries of online usage for children, often stakeholders in this debate: policy writers, researchers and schools. First, at one end of the spectrum I consider arguments that both online and offline friendships are more or less the same; second, at the opposite end of the spectrum I consider that there may be significant differences such that they are very different relationships which may need further, perhaps significant, research attention prior to educational policy creation and suggest that there are encouraging pathways through this quagmire. A third position is derived from taking a mid-point view between the two: that there may be similarities or differences that need to be taken into account. If online friendship is viewed as being of a certain nature (the same, different or somewhat the same), this needs to inform the development of educational policies.
In the next section, I explore recent arguments on Aristotelian friendship and look at how the debates in these previous contributions have been applied to online relationships. In doing so, I point out that the platforms critiqued have moved on from social network sites (such as Facebook) to social media (general) and immersive virtual worlds (e.g. gaming). The following section is in three main parts, devoted to each of the possible positions outlined above and indicating what the implications for the three stakeholders (researchers, policy writers and schools) might be under such conditions based on advice given in a new policy in the UK: Relationships: Education, relationships and sex education (RSE) and health education (DfE, 2019). Finally, I suggest ways in which research, policy and practice might link together.

The possibility of Aristotelian friendship
Why should we care what Aristotle says about friendship? From a purely philosophical perspective, there are two reasons. First, Aristotle in The Nicomachean ethics (Aristotle, 1953) gives one of the earliest full developments of friendship and its relationship to virtue: that friends help us to cultivate certain character traits and having good friends makes it more likely that we will achieve some of our most important goals (given that friends help each other). Aristotelians have always purported that the 'good life' is one that is lived amongst others, for which we need ongoing commitments to a life 'lived together'-genuine involvement in each other's lives in joint activities. Second, most of what he said has withstood the test of time and still rings true in some form today. Much of the theoretical work across the social science disciplines identifies broadly Aristotelian features to friendships without necessarily realising its origins: that it is a voluntary, dyadic relationship; that it is characterised by selfdisclosure, reciprocity and intimacy; that friends typically care for each other and seek each other's good; that friends spend time together in shared activities (Pahl, 2000;Buote et al., 2007). Within philosophy, Aristotle's typology of friendship that falls into those based on usefulness (utility friendships), enjoyment of each other's company in shared engagements (pleasure friendships) and valuing each other based on character (virtue friendships) still proves the most widely used model (Aristotle NE), despite copious attempts to supplant it with alternatives (e.g. Cocking & Kennett, 1998). Whereas the first two types (utility and pleasure) are regarded as somehow 'incomplete', it is the third type (character or virtue) that has attracted most philosophical attention as being 'complete' and the most desirable. Nevertheless, the two lesser forms are regarded as real friendships, but lack many elements that result in a deeper, 486 M. Healy longer-lasting friendship. The overriding presumption of this model is that friendship is always towards the good: that both parties love each other, come to share a similar set of values and commitments and wish each other well, thus giving the relationship significance as a vehicle for self-examination and moral excellence.
For Aristotle, moral character and virtue friendships are intimately intertwined through the 'recognition of the other person's moral goodness' (Cooper, 1999: 315). Because they take considerable time spent together in shared virtuous activity, these friendships are exceedingly rare. Whilst for some the model holds up an unachievable model of friendship, many philosophers argue that a more human interpretation of the general shortcomings can be found in this that would allow us to draw on aspects of the model without commitment to all of Aristotle's conditions (Cooper, 1980;White, 1990;Kapur Badhwar, 1991;Kristj ansson, 2015).
Across disciplines, the importance of friendship has also been linked to both personal and societal well-being. Indeed, a wealth of accumulated evidence consistently highlights the critical role that friendship plays in child development, supporting and promoting important prosocial skills (Damon, 1977;Buhrmester & Furman, 1987;Baumeister & Leary, 1995). Schools are commonly viewed as an important site for the formation and development of friendship, based on both proximity and time spent together, such that they have come under the microscope as part of the pressure for continuous improvement (Ladd, 1990;Day, 1996). Compelling evidence seems to point to the fact that children who have positive friendships enjoy better mental health (Ueno, 2005;Holder &Coleman, 2015), and that this in turn can be linked to academic achievement (Fenzel & Blyth, 1986;Day, 1996;Wentzel & Ramani, 2016).
The bone of contention (at least for philosophers) is not over the possible existence of virtual friendships per se (which is more or less now accepted), but over the possibility of achieving this third type online: deep (or Aristotelian virtue/character) friendship. On the one side are those who argue that online friendships cannot be genuine virtue friendships and thus deny that there is a moral equivalence between those friendships on and offline; on the other side are those who argue that online friendships may meet the requirements for virtue friendships when understood in particular (usually restricted) ways (e.g. as a form of 'shared life'). Much of the growing literature in this area tends to cover three distinct categories to which we now briefly turn: social network sites (Cocking & Matthews, 2000); social media (Elder, 2014;Kristj ansson, 2019); and online gaming (Nicholas, 2014).
Nearly two decades ago, philosophers Dean Cocking and Steve Matthews concluded that what we generally understand in philosophy as deep (or virtue) friendship was unlikely to evolve solely from the text-based communication associated with 'virtual friendships' (Cocking & Matthews, 2000). Much of the relevant debate has expanded on Cocking and Matthews' (2000) arguments claiming that text-based friendships, such as those conducted on an SNS (social network site), would inevitably filter, limit and distort the aspects of the self being portrayed to such an extent that deep companionate friendship (when understood as Aristotelian 'virtue friendship') could not develop. This ability to manipulate and distort one's self-presentation by hiding information, personality traits or values (which would normally emerge in everyday interaction and activities between real-life friends) via a medium that enabled the creation of an 'idealised self' was considered to disqualify such Educating for online friendship 487 relationships from reaching the depth of mutual recognition needed for such a relationship. Moreover, Cocking et al. (2012) continues to suggest that the online 'self' may still not represent our 'true' self at all, in that 'various aspects of self about which we do not approve, or that we think are not notable or that we simply do not notice are not revealed or are distorted ' (p. 180).
Whilst more recent developments in the usage of such technology (particularly the growth of SNSs) have caused the authors to re-evaluate this stance (Cocking et al., 2012), such philosophical debates over the possibility of virtue friendship in a virtual world continue. On the one side stand philosophers such as Fr€ oding and Peterson who argue that online relationships could not be classed as genuine virtue friendships, but only as a lower form of friendship from an Aristotelian perspective, lacking one of the key conditions needed (Fr€ oding & Peterson, 2012). Fr€ oding and Peterson argue that Aristotelian friendship needs to be mutually recognised, to take place between two adult humans of equal standing and that such friends need to spend time together engaged in shared activities, all of which could be met online. This leads them to maintain (as do Cocking et al.) that Aristotelian friendship is founded in virtuous behaviour by both parties. However, they argue, the love and admiration each person feels for the other in Aristotelian deep friendship is based on the virtues recognised in each other; this in turn requires honesty and truthtelling about who you really are that may be absent or withheld when online. Failure to have this means one party could have less than full knowledge about the other, such that the foundations of the friendship may prove to be illusory. This, they hold, makes the promise of SNSs 'ring hollow' as they would not be able to support deep friendship without any real-life contact. 4 In contrast, more nuanced writings have emerged from philosophers such as Elder (2014). Despite being in broad agreement over the nature of Aristotelian friendship, Elder suggests that support for the many unique attributes of character friendship might be found in broader understandings of online communication: for example, that social media has the potential to support Aristotelian character friendship when understood as a 'shared human life' (Elder, 2014). In many cases, those we are most in touch with online are the same people we are friends with in everyday life-and studies do tend to support this (e.g. Tang, 2010;Amichai-Hamburger et al., 2013). Individuals can 'share' their lives by sharing activities with their friends-in this case, sharing conversations, pictures, videos and playing games together online is part of creating this friendship-hence 'virtual friends' no longer need to be geographically present in order to share each other's lives. She thus denies that the online components of friendship are at a significant disadvantage to the offline components (Elder, 2014). Indeed, she further argues that such online friendships may simply function as a 'stop-gap'-a way of keeping in touch-between real-life meetings.
This lively debate by philosophers (such as Cocking et al. and Elder) has moved from concerns around SNSs to social media-and now moves to immersive virtual worlds with their emphasis on joint shared experiences and activities as a possible way in which deep friendship may be developed. The emerging view is that these may avoid the problems of the cue-restricted social media such as SNSs, micro-blogging, etc. with their dependency on text-based communication and thus will not be as accommodating at hiding or distorting information (Welles et al., 2014). This new line of argument suggests that online gaming is a major way that people play games together, maintain social relationships and thus build the possibility of a 'shared life' (Liu, 2010;B€ ulow & Felix, 2016). Spending time together in joint activities is an essential part of friendship in this view (an Aristotelian expectation of deep friendship). In another recent paper, Nicholas argues that the social interaction of the game might perform some of the same functions as other real-life encounters and suggests that gaming could be seen as a way of trying to find friends and that the bonds formed in playing characters often mirrors real life, where they have to depend on each other to achieve (Nicholas, 2014). When it comes to arguments as to whether or not such sites can offer the type of 'loving involvement' associated with friendship as described by philosophers (Annas, 1977;Cooper, 1977;Pangle, 2003), studies finding evidence of online activity resulting in offline support are inconclusive (Trepte et al., 2012 argue for; Pollet et al., 2011 argue against). How might these philosophical arguments inform the development of educational policies?

Possibilities for online friendship
Although the internet was initially seen as revolutionising our access to information and data, SNSs have rapidly become its major use (Cocking et al., 2012). Started initially as a means for Harvard students to keep in touch after graduation, Facebook has become one of the largest social networking online communities, used by millions worldwide and is usually one of the first SNSs used by children. The influential PEW Research Centre claims that 57% of American adults and 73% of those aged from 12 to 17 use Facebook, with many (64%) visiting the site on a daily basis. 5 For this reason, I will be taking the features of Facebook as a generic example of SNSs for illustrative purposes. 6 Feeling socially connected to others is perhaps a human need, and Facebook in particular would seem to have tapped into this need for connectedness beyond its original purpose-but not without controversy. Its advent was claimed by some to have modified the term 'friend' so that the term, whilst still commonly used as a noun, has become increasingly used as a verb: 'to friend' someone, based on the online practice of adding 'friends' to one's network. For some, this was seen as a debasement of friendship, which would inevitably affect the value of the relationship in real life by replacing it with something more trivial (Amichai-Hamburger et al., 2013). Yet the concept 'friend' has always been subject to change-from the Greek exemplars of 'philia' (covering familial relationships, civic relationships as well as friendships of many years) to more modern usage in social relationships, quasi-kinship and general peer relationships. So adaptations in usage need not themselves necessitate that the traditional understanding of friendship is now redundant, nor that the importance of friendship for social development has lessened in any way (Amichai-Hamburger et al., 2013). 7 Despite being in broad agreement with the overall arguments over the possibility of character friendship online (given particular conditions), none of this work considers how this might look in the world of children, nor does it offer a framework for introducing children to such relationships. Nevertheless, this literature can act as a catalyst for examining the sections on friendship found in the new Relationships policy (DfE, Educating for online friendship 489 2019) and teasing out how this might look in schools. In what follows, I outline three different ways this philosophical literature could have implications for education and indicate the types of policy that each might entail. I will contend that there are three distinct stakeholders here that can be identified as having a significant role to play in what we should do: the schools themselves, educational researchers and policy writers. Furthermore, I will suggest that each of these approaches can be linked to a responsibility for these stakeholders.
Possibility 1: Friendship on and offline is more or less the same The first point to note is that in the virtual world, a 'friend' relationship is said to exist when a request to another on the site is accepted by the press of a button. The relationship has to be agreed by both parties (the initiator of the request and the recipient who accepts, akin to the mutual regard of Aristotle). Yet in the real world, friendship is far more complicated, requiring considerable time spent together in shared or joint activities (Cooper, 1980)-a key feature of Aristotelian character friendship. In accepting a friend, one gives permission for them to read and respond to photos, videos and comments that have been uploaded onto the 'wall'. This gives the opportunity for dialogue and discussion at a simple level and the beginnings of a shared venture (all Aristotelian features of friendship).
Research seems to indicate that most people request friendship on SNSs with people they have met in real life (friendship maintenance), with far fewer seeking to initiate friendships in this way (many combine real life with online life-blended relationships), and even fewer initiate and maintain their relationship solely online (Amichai-Hamburger & Ben-Artzi, 2003;Tang, 2010). As most online relationships are with people already known, they might be more accurately viewed as being about 'absence' (Holland & Harpin, 2008). This would seem to imply that remote relationships (conducted online, by phone, email or snail mail) do not necessarily replace offline friendships, but merely provide ways to maintain and/or extend existing ones.
This might at first glance draw us towards arguments that friendship on or offline must be more or less the same. Indeed, arguments in favour of such a supposition are commonly made: for example, both are voluntary and chosen; members share privileged information with each other (self-disclosure); both may have some expectations of trust and confidentiality around communications under certain circumstances (e.g. private messaging); they may 'spend time together' (sharing interests, gaming, etc.) and engage in shared activities-again, features of Aristotelian friendship (see Amichai-Hamburger et al., 2013;Elder, 2014;Welles et al., 2014).
This first possibility is obviously within the remit of schools-they have the primary face-to-face challenge of working with children, keeping them safe and supporting families and communities with the ongoing development of character, often through a taught programme (such as citizenship) and the school ethos. But what sort of educational policy might schools need in this model? The first position would best be met by promoting a 'person-centred' approach, 8 based on some form of character or moral education (sometimes referred to as PSHE-Personal, Social, Health and Economic Education; DfE, 2015), perhaps with some attention to the dangers of the mode of communication (i.e. e-safety and 'stranger danger'). Indeed, the new 490 M. Healy Relationships policy explicitly states: 'pupils should be taught how to take turns, how to treat each other with kindness, consideration and respect, the importance of honesty and truthfulness, permission seeking and giving. . .' (DfE, 2019: 19). In many ways, this builds on the advice given in the policy for Computing 6 years earlier, where it was dealt with in two sentences and not elaborated on: 'use technology safely, respectfully and responsibly; recognise acceptable/unacceptable behaviour' (DfE, 2013: 2). Similarly, the current advice is that the same principles of positive relationships should also apply online (DfE, 2019: 19-20;Long, 2018): 'The principles of positive relationships also apply online especially as, by the end of primary school, many children will already be using the internet'. This would then justify accentuating the place of personal attributes (respect, responsibility, etc.) as we might hope that some form of cross-over would take place. This would fit in well with other, similar social development programmes currently aimed at the youngest children (DfE, 2015, 2017)-and perhaps could be used as a reason to justify a firmer place for PSHE on the curriculum. This makes the linkage to the Aristotelian view of friendship even more apposite: for an Aristotelian, friendship is about the development of virtuous character and it is clear from the above discussion that much of this links directly to such premises. But we should also bear in mind that not all friendships are ones of virtue: vice friendships can and do exist whereby two (or more) are drawn together not by virtue, but by human vices (Healy, 2011). So there is always the possibility of bullying in all communities-online and in real life.
But we have to be careful before leaping straight to this conclusion and making this our preferred way of underpinning policy. As yet, we lack sufficient research to draw accurate conclusions as to whether such attributes of character are the same both on and offline, even for adults: that those who are likely to be honest (for example) in one domain are likely to be honest in another-and there are good reasons to doubt this. Research has shown that the anonymity associated with online interactions can promote decidedly unvirtuous behaviour in some that they would find abhorrent in real life, such as trolling and abusive threats (Jones et al., 2013). Even if we use the same word in both contexts, it does not automatically carry the same meaning or connotation, and this factor can sometimes lead us astray. But it also highlights more practical needs for schools, and suggests a need for further staff development in this area to ensure staff have both the experience and the knowledge to undertake this task. So, whilst we might be tempted to see online and offline friendships as being more or less the same, we may need to consider such assumptions more carefully after further empirical research.
Possibility 2: There are similarities between friendship on and offline, but there may be differences that need accommodating Whilst many philosophers would accept that the two lesser forms of Aristotelian friendship (pleasure and utility) can align with some attributes found in online relationships (as considered in the previous section), the third element (character/virtue) is still seen as problematic-in other words, it needs to be taken into account. But what exactly are we looking at here? Elder (2014) offers yet another perspective, that here we have to return to Aristotelian principles: what distinguishes the highest Educating for online friendship 491 category of friendship from the others is that the traits of usefulness and pleasurability do not exhaust the value of the friendship. Character friends have value over and above these traits because of the sharing of their lives together in a special way, each becoming 'another self'. The sharing of personal thoughts and feelings, the depth of knowledge each develops about the other over time, is argued to build an intimacy that validates each friend's feelings of personal worth in addition to the self-knowledge and self-perception gained. But more than this, we learn to cherish the other (their needs and wants, their ambitions and desires) as they truly are, so that they become as important as our own needs and wants and we act to advance our friend's flourishing as we do our own. To achieve this, we need a high level of honesty and openness to one another about who we are and what we truly value. Now it is highly unlikely that many children will reach this level of moral discernment, and so we need to think about what is possible at particular ages. Here there would be a need for involvement by educational researchers as well.
What counts as a 'shared life' for children is not easily identified. Undoubtedly, sharing activities is not just about the pleasure gained, nor the utility of a playmate: in sharing, we create a space in which to learn about and appraise the character of the other. Do they play fairly? Do they cheat? Do they share? Are they a good loser? Again, this ties in with our interest in character. But what it is we 'share' can itself be contentious. We are becoming increasingly open in the way in which we publicise our persona online, and perhaps less concerned at the multiple ways in which others can access and process data about us that we willingly put online. Admittedly, social norms and expectations vary across fields of study, and shifting social boundaries between the public and the private are continually under negotiation. West et al. (2009), for example, claim that students, having grown up with little to no experience of life prior to being constantly online, seem to have developed a somewhat 'hazy' understanding of the difference between public and private. This becomes particularly an issue when we start to identify the dangers this may put children and younger adolescents in, who may have even less understanding of the dangers this may result in for themselves. Or, as policy puts it: 'the distinction between online and offline may be less distinct to children and that this may leave many at risk without appropriate education' (as recognised by DfE, 2019: 9-10). This could itself be a good reason for policy to highlight both staying safe online and friendship. If this second position is adopted, we might then need further advice in using it appropriately to deal with the differences between on and offline relationships. And this is where policy stakeholders have a role to play. It should not be beyond them to simply update/replace outdated policy advice in the light of changes and developments, or to seek coherence both within policy documents and between such documents. For example, this new Relationships policy (DfE, 2019) is still in draft form and as such could then include a little more on online friendships (and not just the catchall 'relationships') and the changing nature of social media platform usage. But policy writers do not work in a 'vacuum' and are themselves dependent on the work of school staff and researchers. For example, in recent years an important theme amongst philosophers of education has centred on how philosophy can be brought to illuminate the landscape of policy and practice, and many philosophers may have direct expertise in this area (Arthur et al., 2016;Bridges, 2017;Curren, 2017;492 M. Healy Webster & Pring, 2018;Orchard & Davids, 2019;White, 2019;Biesta, 2020). Whilst many philosophers of education are now or have been teachers, as school staff have to enact such policy, it makes sense to not only bring them into such conversations, but also listen seriously and take account of their particular expertise. Unless policy can be enacted, it remains empty words.
Possibility 3: There are significant differences between friendship on and offline such that they are very different relationships To consider the third possibility, we need to return to the original concerns raised over online friendship: that online relationships are somehow more dangerous or less morally valuable than those lived offline (e.g. Fr€ oding & Peterson, 2012;B€ ulow & Felix, 2016). In favour of the similarity between the two relationships (online and offline) are arguments that friends influence and shape our outlook, that this validates our feelings of self-worth and thus supports us in gaining the self-knowledge to live 'a life of virtue'. Those to whom we entrust this role (usually our dearest friends) hold a privileged perspective that may not be open to others in the same way (sometimes referred to as the Aristotelian 'mirror' argument). Nevertheless, the problem seems to remain that unless one controls the membership of the SNS, it is highly unlikely that sufficient members will be made up entirely of friends that care enough about us to always give the communication the emotional or moral significance we expect in deep friendship (hence indirectly supporting Cocking and Matthews' original arguments). But there is at least the theoretical possibility that some may do so. 9 Yet even in cases where no blended relationship exists (combining online and offline life), I do not want to completely rule out the possibility that even completely online friendships may reach this depth of intimacy and self-disclosure. Take, for example, an adolescent agonising over perhaps a gender-identification or disability issue who might not have access to others who share the same experiences as they have, or access to relevant supportive services. It is quite feasible that they might join an SNS group specifically formed by others in such a position and, over time and given the possible mutual disclosure and support offered, some may come to know each other at a very deep level, allowing the relevant intimacy to develop, and to care deeply about each other's well-being, typified by all the attributes we generally expect in friendship-even if they never meet. Indeed, some people may find such purely online friendships to be as close and fulfilling as any of those in real life.
This third possibility reminds us of the importance of accurate research in this area. As Kristj ansson (2019) controversially suggests, there is a dearth of research on online friendship that can really be depended on. I want to go further and argue that this is even more of an issue for research undertaken with children using social media. Now there can undoubtedly be significant ethical issues involved in undertaking sensitive research in such areas (or even with having access to child subjects), and research dates very quickly in this area as children move between platforms as they age/follow trends, but the lack of appropriate research impacts on our ability to make useful policy for schools to enact. Whilst this seems, on the surface, an issue mainly for researchers, it becomes an issue for school staff (and pupils) when the policy 'fails' Educating for online friendship 493 to address the right issues in the classroom. 10 But more importantly than this, it raises questions about how we understand the world and how we use metaphorical models to respond to this 'newness'. How we interpret the world, the models we use, forms a crucial link in understanding what social life is like and how things fit together. According to theorists such as George Lakoff and Mark Johnson, much of our social reasoning makes use of interlocking systems of concepts (Lakoff & Johnson, 2003).
In new situations, we need a 'new vocabulary' (Healy, 2013), so in calling online contacts 'friends' we then carry over to the new concept (online relationships) all of the images and cumulative thoughts from our personal real-life relationships. In truth, they may turn out to be very different relationships indeed. The next section makes some policy suggestions arising from this discussion.
Conclusion: moving towards policy I examined three distinct positions. (1) That friendships on and offline are more or less the same thing-I conclude that we cannot depend on such a supposition; but if we did, it would require a person-centred approach to fulfil this view. (2) That there are similarities between friendships on and offline, but that there may need to be some modification. Thus, what counts as a 'shared life' for children needs philosophical attention.
(3) That friendships on and offline are very different relationships and that we need further research in this area. In each case, I have drawn out some of the implications for education and this has entailed identifying particular stakeholders who might find these approaches useful.
We have to be careful in assuming that schools can address all social problems and perhaps accept that schools may only have a limited or supporting role to play here regarding the dangers highlighted at the beginning. Brighouse (2005) claims that schools have become increasingly expected to make up for the failure of other social institutions, yet the very idea of the 'educationalisation of social problems' (pushing social problems into schools as a way of addressing societal problems) is increasingly seen as problematic and critiqued (Smeyers & Depaepe, 2009). For some, what is being suggested here may also have a similar tendency: that friendship is seen as part of private life and therefore should not be subject to public policy. In answer to this, I want to make two points. First, philia as used by the ancients covered both personal and public relationships (from friendship to civic friendship)-so friendship has always had this double-faced aspect to it, both private and public. Second, and perhaps more persuasively, much of what we might call 'private life' is also becoming subject to public policy-domestic abuse, child care/neglect, etc. That public policy has expanded to cover particular rights within the private domain does not always infantilise those involved, as Smeyers and Depaepe (2009) suggest, but neither can it solve all problems.
The virtual world is increasingly an important context for friendship and social development, and the ways in which we conduct our relationships (particularly our communication with others) have undoubtedly changed in recent decades. The rapidly evolving nature of 'social media' means we can be permanently 'in touch' with the world and the world 'in touch' with us in ways undreamt of in the past. As friendship plays such an important part in children's social development, it is unsurprising 494 M. Healy that it is so often studied purely in terms of the usefulness and the 'outcomes' generated: social support, social skills, behavioural influence, conflict resolution and issues of personal identity reinforcement (Sullivan, 1953;Buhrmester & Furman, 1987;Newcomb & Bagwell, 1995). Nevertheless, there are other reasons for studying friendship-it is also about what we ultimately value and what makes life flourish. And this is where the work of Aristotle can make a contribution. Aristotle's theory of the good life as presented in The Nicomachean ethics (Aristotle, 1953) sees friendship as a necessary element of what makes life worth living, and it is perhaps a tribute to the longevity of Aristotelian arguments on friendship that much of this contemporary debate is still so closely tied in with his work.
One result of this has to be to treat the issue of practice, policy and research as intertwined, with each feeding into, deepening and challenging the other areas (Bridges & Watts, 2008;Smeyers, 2008). Some final points are worth reiterating and emphasising. First, friendship is not a simple concept and increases in complexity when positioned online. Given the social importance of friendship as a significant peer relationship, and given that schools are often required to make provision for personal and social education, schools still often struggle to keep up with the fast-changing nature of social media usage. So while it is to be commended that the writers of the new policy in the UK (DfE, 2019) recognise that not all relationships for our 'digital natives' are lived offline as a first step towards addressing this issue, schools may need further advice to implement or deliver this into pedagogical practice. Second, most of the children in any one class will have been interacting with technology in one form or another from early childhood, whether watching TV shows or playing games on iPads/tablets, using smartphones or similar. Many of these children will know more about how these things work than their parents, and some of them will be using them unsupervised, hence there is a need to address safety issues both with children and their caregivers. This seems to suggest that children and their caregivers also need to be included as possible stakeholders in some of these conversations. Third, ensuring children are safe and addressing the character attributes and dispositions needed for civic living/self-protection (e.g. trust, honesty, etc.) is a significant responsibility for all adults, no matter what their stakeholder status might be. Fourth, as yet there has been little to no cross-over between empirical research into the virtual world and the world of children, with one or two exceptions (Kristj ansson, 2019). Yet this is an area where philosophers of education may be able to offer their particular skills (e.g. helpful clarification around concepts used, highlighting contradictions, illuminating conceptual distinctions, to name but a few). Finally, if policy or practice is to be formed around the development of virtues, we could do far worse than turn to Aristotle and contemporary writers in that tradition who remind us that we are 'social animals' and that friendship is crucial for well-being and the development of virtue-and not just academic success.