Incommensurable Ethics

The unsuitability for anthropology of formal processes of ethical review and administration employing concepts and schema drawn from the traditions of human subject research have been highlighted in an increasingly voluminous and insistent literature, which has drawn attention inter alia to the impossibility of ‘doing no harm’ in the process of ethnography (Bell 2014: 512), the problems of ethical reflection decontextualised from the realities of the field (Sleeboom-Faulkner et al 2017) and the general inevitability of transgressing against one guideline or another in the course of fieldwork (Verhallen 2016). Informed consent, as Kirsten Bell has pointed out in her argument against the commensurability of the concept with anthropology, has often avoided the brunt of these attacks (‘after all, how can one be against informed consent?’; 2014: 512) but has nonetheless been the subject of recurrent discussion and challenge – not necessarily to the worthiness of the concept so much as the possibility of its useful application.

These challenges apply, in various degrees, to the majority of contexts outside of anthropology in which informed consent is generally sought – one may question, for example, the degree to which one may be genuinely informed about complex medical procedures, or be able to give meaningful consent to researchers investigating an illness one currently has (see Corrigan 2003). However, I argue that the particularities of anthropological practice make it ethically distinct from cognate disciplines: through its inherent commitment to holism and context, the constraints on potential outcomes that allow more structured disciplines to predict the potential ethical quandaries involved in the course of research are made impossible. I will return to the case for anthropology's disciplinary exceptionalism later, when I discuss the manner in which we can look to the logic of the practice of anthropology itself for ethical guidance. For now, however, it is enough to emphasise that the openness and inductiveness of the discipline's methodologies underlie the particular salience of the most penetrating critiques of the universality of informed consent as an ethical principle when it is applied to anthropology.

The disciplinary commitment to context in ethnographic description and theory has both methodological and epistemological consequences for this. The anthropologist, as Barrie Thorne (1980; see also Murphy and Dingwall 2007) has argued, will often find that their professional role as a researcher fades into the background in the course of fieldwork. Indeed, the elusive emic perspective can be supposed to demand such a fading. In any case, the expectation of long-term participant-observation is – as all new PhDs are warned – that no research question survives contact with the field (see Lederman 2006); to suppose otherwise is hubris. This is in many ways the great asset of the discipline, as it allows for emergent and innovative threads of inquiry – but it also precludes the possibility of any comprehensive consideration of the ethical field ex ante. Even if a researcher should be in a position to secure from participants the formal written consent that many codes of ethics demand – a requirement that, for various reasons, is often infeasible (Wynn and Israel 2018) – the blanket notion of informed consent to participation in research secured at the outset must be discarded from the first, on the basis that no anthropologist can themselves be assumed to be fully informed at that point, let alone in a position to bring others up to speed.

My concern here is not, per se, with the unjustified hindrances that ethics administration has put in the way of ethnographic research with the imposition of such clearly impractical standards (see Sleeboom-Faulkner et al 2017), nor with the thinness of informed consent as a concept. Rather, it is with what should serve as an ethical lodestar once such criticisms have been acknowledged and accepted. Tessa Verhallen (2016), in her exploration of the broad inapplicability of codes of ethics in the ethically fraught context of child protection, has argued for a contextual ethics that prioritises the interests of research participants and once again focuses ethical questions on the (inevitably subjective) professional discretion of the researcher. Her critique of the concept of informed consent is comprehensive, but she focuses particularly on power relations, the interests of her participants and the validity of her data to justify her ‘semi-covert’ methodology, through which she seeks neither to reach an impossible level of informed consent, nor to offer those with different levels of power the same degree of consent, choosing rather to privilege the vulnerable.

As I discuss in the following section, Verhallen's ultimate conclusion – that a flexible ethics emerging from the context of the fieldsite and constructed on the back of trust, rapport and enduring relationships with participants was more practical and appropriate than formal codes of ethics – was very similar to my own in the ethically perilous conditions of Xinjiang. So, too, was her conclusion that obtaining the consent of the powerful is ‘unfavourable’ (2016: 463) if doing so harms the powerless. For Verhallen, the ‘contingent arena’ of the field, in which various conflicting ethical dilemmas arise, is such that no one formula can guide the researcher towards ethical practice – a position that necessitates re-emphasising of the anthropologist's discretion over normative codes.

Here, I hope to build on this emphasis on discretion, which I argue must be the keystone, explicitly or not, for any form of guidance – in the context of a situation in which the enduring relationships characteristic of anthropological fieldwork have been forcibly stripped away and have no imminent prospect of reconstruction. Indeed, in the contemporary context of Xinjiang, which I outline below, attempting to re- establish such relationships would put my former research participants in significant personal danger and, as such, would in almost all frames of reference be clearly unethical. This severance demarks, perhaps with a finality only surpassed by the death of those involved in a piece of research, the difference between the ethics of the field and those of representation. Paradoxically, however – and in contrast to the distinction that Quetzil Castañeda draws between ethnographic fieldwork and representation on the basis that the former relates to ‘lived-in interactions’ (2006: 122) and the latter to questions of abstraction, reification and othering – this same context ties representation intimately and immediately to the lives of those being represented. Like those who have given consent to research prior to death, most of my erstwhile participants are effectively beyond communication. However, they remain potentially at risk. The circumstances in which consent was given and data collected have been upended, such that it is reasonable to question if consent would ever have been given by many of my participants had they had any inkling of the change in conditions that would commence almost immediately after I had left the field.

Access and Ethics

This article is not intended to provide another ethnographic example of the inadequacy of codes of ethics for the practice of anthropology: that path is well-trod already. It was, in fact, necessary for me to draw heavily on critiques of normative ethical administration prior to gaining ethical approval for fieldwork in the first place. The conditions in Xinjiang in 2012, as I prepared to begin my research, were already fraught, if not comparable to those in more recent years. Strict adherence to the letter of professional codes would have rendered my research impossible – and, indeed, I am aware of other researchers who have proposed similar work and been denied approval. I was fortunate that the ethics review board that considered my proposal had enough expert anthropologists on it that the necessary flexibility was recognised.

In broad strokes, the conditions at that time in Xinjiang that necessitated such flexibility were the result of state oppression of minorities in that region, a long history of ethnic tension with recent violent incidents, a disputed political past and future, and significant efforts by the state to control information about the conditions of life there. Researchers had, to varying degrees, experienced difficulties in gaining access to fieldsites and contacts in Xinjiang since communist China opened up in the late 1970s and 1980s. While there had been periods in which state controls had been relatively relaxed, they had never been so lax that researchers could broach all topics without significant caution regarding who they might be talking to, or who might be listening. The risk was not to the researchers themselves – beyond the possibility of losing access to the region, or China as a whole – but to their participants and other interlocutors.

Generally speaking, social researchers have adopted two approaches to securing access to Xinjiang, both of which come with methodological limitations. The first is to negotiate research access with the state, through association with a Chinese research institution. While this has provided some security for researchers, it also immediately alerts authorities to their presence, resulting in restrictions on the topics that can be explored, interference in the research process and the coaching of participants (Joniak-Lüthi 2016; Smith 2002). In any case, access of this kind was only granted for topics not considered to be too sensitive – something of a narrow window in Xinjiang.

The second approach involves covert or semi-covert research and thus begins with a violation of standard demands of transparency and informed consent. It also contravenes the expectation that researchers maintain a particular kind of respectful relationship with the authorities of the state in which they carry out their research.¹ Active research participants are not misled by the researcher, but potential participants are not informed of what the researcher is doing until a relationship of trust has been built up between the two. It is thus an approach that explicitly involves a conscious (and conscientious) application of the kind of adaptive relationship with transparency that Annelies Moors (2019), among others, has argued is implicitly an element of all ethnographic fieldwork. However, it also requires – for the protection of participants – that the state and untrusted individuals be explicitly misled about the reason for the researcher's presence in the region. Thus, the researcher is constrained to not only mislead by omission in the day-to-day course of their research but to actively lie as an ethical necessity.

There is no sharp distinction between these approaches, not least because of the rapidly changeable nature of political sensitivity in Xinjiang. Those researchers given permission to study politically acceptable topics have inevitably observed the realities of the oppression of the region's minorities, not least because many among them are so eager to discuss precisely the things that are most dangerous. All this underscores the reality that the ethical principle given perhaps the most weight in normative codes – do no harm – would preclude almost all research in Xinjiang if it were held to be paramount over all other considerations.

I cannot here outline the full complexities that arise from these issues. What is important for this discussion is that, under such circumstances, the constant renegotiation of consent was not an abstract principle returned to iteratively but a practical reality emphasised in the first instance by my informants. Part of this was the gradual movement towards more risky topics – participants and friends would, as we built trust, gradually change the way they talked about such subjects as job prospects, intermarriage, Han people and treatment by the state. While my presence was an inevitable risk to those I worked with, this careful attenuation of caution and defensive misrepresentations of personal perspectives allowed them to manage – in their own terms – their exposure to the most proximate and severe risks. In recognition of the fact that decision-making (ethical or otherwise) works best when well informed, and that my informants were far better placed to identify, in the immediate sense, what was dangerous (on this point, see Hodge 2013; Kovats-Bernat 2002), my own steering of unstructured conversations largely followed from these cues.

The information on which ethical decision-making must draw in the field is not available before fieldwork begins – indeed, only through the process of fieldwork and interaction with the people whom one has gone to study can it be obtained. It is on this basis that advocates of situated ethics have argued that the focus of consideration for ethical ethnography should be the ethnographic reality ‘on the ground’ (Verhallen 2016: 470). What was available in advance of fieldwork, however, was a blunt appreciation of relative power relationships in the region, and a sense of the sorts of methods that would be necessary to negotiate those relationships in order to approach that ground in the first place. While this appreciation would be sharpened by the processes of fieldwork itself, it nonetheless represents a wider logic of ethical consideration which, for the anthropologist, is necessarily both prior and external to fieldwork. I will return to this logic shortly, to make the claim that it is – in the last instance – the ultimate guide for anthropological ethics. First, however, I will outline how the shifting of circumstances in Xinjiang has potentially delegitimated the consent given during my fieldwork, and precluded any further negotiation.

Ethics Out of Context

The gradual opening up around sensitive topics that I outlined in the previous section took place with reference to various different sorts of awareness. While my research participants were far better informed than I was as to the local dangers, they were relatively uninformed about the potential uses to which their information might be put. My engagement with them was, in all cases, necessarily begun without full candour. My approach to fieldwork access was the second of those discussed above – I made only my active participants aware of the fact and nature of my research, and only at the point that I asked them to engage in the research. To others, including authority figures, I justified my presence variously with reference to tourism, work or accompanying a family member who had enrolled at a local university – explanations that allowed me to travel in the region and to visit various educational institutions. However, my participants’ lack of awareness was not limited to the covert elements of my fieldwork but related also to a common lack of relevant experience by which to make sense of academic publishing or the technical measures I took to protect their data. However – and this is the flipside of an ethics built on enduring relationships – the information that was given to me was chosen on the basis of my participants’ active judgement regarding not only my trustworthiness but their conception of their own level of risk. Their consent, then, must be understood as definitive only insofar as those conditions remained – which they do not.

My first case is that of Anargul, a young Uyghur woman. This is a pseudonym: her real name has never been a part of my notes. She had requested that I not record her name, but I would not have done so in any case; anonymisation from the outset was a method employed to reduce the danger of my notes being taken from me and read before I could secure them. Like all anonymisation, however, this was partial – her transcript contains details that could conceivably be used to identify her, for example a distinctive educational pathway; her father's profession; her age; her home town. These were relevant details for my research and by preserving them I was balancing – as I continue to do now – her safety against the integrity and value of the output. This balance is an unavoidable element of all research into sensitive subjects.

My conversations with Anargul started with her home life. The basic facts of her experience would later become central to the ethical dilemma around using her data. She was a minkaohan – a minority – who had been educated in a Chinese school. She thought Mandarin had become her stronger language – she only read Mandarin books – but her parents feared that the Uyghur language may die, and she only spoke Uyghur at home because otherwise her father would be angry. She had begun to sympathise with this and now considered that a minority-language school would have been the better choice. She was in her early twenties and planning to find a graduate of a minority school to be her husband, so that at least one of them could pass Uyghur language skills and culture on to their children.

These were familiar topics, experiences repeated time and again in interviews with young Uyghur adults. She considered our discussion relatively sensitive: we talked, for example, about her feelings about wearing a headscarf, her difficulties with racism in Inner China, her dismay that the minority-language schools that she wished she had attended were being closed down. But her limits were very clear: when I asked her why the minority schools had been closed down, she ended our discussion of education. ‘I don't want to answer why they closed [minority] school[s]’, she said, ‘I won't tell you why.’

Her refusal implicitly referenced the suspicion, shared by many Uyghur people, that the schools were being closed as part of a concerted political effort to undermine Uyghur culture. By not explicitly giving a more politically acceptable answer, she was expressing this while avoiding the significant risk of direct criticism of the state. Her consent, then, can be understood – at least in terms of the proximate dangers of which she was aware – to extend only up to a certain level of risk, to which she was acutely sensitive and I somewhat less so.

My second example comes from Erkin, a man whose considerable success was built on just this kind of sensitivity. His business was one that was allowed to continue only through his canny navigation of personal connections and cautious decision- making. A number of his business's previous endeavours had been cut short either by the actions of the state or by his own efforts to pre-empt the state's involvement. A more worldly individual than Anargul, Erkin had a generally well-informed conception of the nature of academic research and the potential risks it could introduce beyond the immediate.

The motivations of research participants are invariably complex, and often in- clude relations of reciprocity, politeness, friendship and trust in addition to more self- interested goals (Guillemin et al 2016). However, among my participants the desire to talk about the political circumstances in Xinjiang was, once initial fears had been overcome, often explicit. Certainly, Erkin stood to gain little from allowing himself to be interviewed, and had a great deal to lose, but chose to do so anyway. Like Anargul, he explicitly curated his discussion with me, avoiding some questions entirely and circumventing others. Similarly, he was willing to cross certain boundaries – claiming, for example, that government literacy statistics were fabricated, and providing examples to support this, or suggesting that the Uyghur language was at risk – but not others. When I asked him if parents were unhappy with the new school system, for example, he simply said that the question was too sensitive. Being better informed than Anargul about the potential risks involved with academic publishing, he also asked that I would share anything I wrote with him before I published, so that he could clarify things and warn me if I was taking unnecessary risks.

Even more than with Anargul, Erkin was difficult to anonymise. I had to be extremely cautious in how I used the information he provided for my doctoral thesis, as to specify almost anything about him would have risked immediately exposing him to the state. However, the circumstances in which I acquired that information have since changed, and in a manner more profound than the usual shifting of conditions that any anthropologist might expect once they've left the field. Power relations between participants, anthropologist and state have all been changed fundamentally.

In 2017, three years after I returned from the field, China's long-running security campaigns in Xinjiang transformed into more generalised processes of curtailing minority customs, language and religion, and ‘reforming’ minority people to be primarily loyal to the state (for an overview, see Roberts 2020). The re-education programme by which the state has attempted to achieve these ends has seen the extrajudicial incarceration of around a million people (Zenz 2019). The criteria by which individuals have been selected for incarceration is neither clear nor consistent, but involvement in Uyghur cultural activities, avoiding the use of the Chinese language and practising Islam have all been reasons for arrest. In addition to the re-education programme, Xinjiang has been heavily securitised in the period since 2017 (Smith Finley 2019). At various times, checkpoints have multiplied and appeared on pedestrian streets; mobile phones have been fitted with surveillance software; facial recognition cameras have been installed across cities; and so-called ‘convenience police stations’ have densely packed every neighbourhood. Under such conditions, any suspicions can put Uyghur people at risk. Like other researchers of Xinjiang, I lost contact with many of my erstwhile research participants as a result of this process: contact with a foreigner could result in incarceration, and both phones and messaging apps are monitored. Even where communication may be technically possible (and for many, it is not), and where both parties would be willing, it is often impossible to initiate because neither can know whether the other would consent, and cannot ethically risk doing so unilaterally (for an example, see McMurray 2020).

Beyond this severance of relationships, however, was a more fundamental severance of the information I had gathered from the considered levels of risk in which it had been obtained. Anargul's clear sense of where she drew the line in terms of what she was willing to discuss is no longer meaningful in light of Xinjiang's new reality. If I imagine what she would consent to should I meet her today, I must conclude that she would not speak to me at all. Had she known what Xinjiang would be like today, I do not think she would have spoken to me then. Some of the topics that she was willing to hazard, on the basis of their being of lesser risk, relate to opinions and practices that now number specifically among those being targeted by the state. Her father's practice of enforcing the Uyghur tongue in the house, for example, is one of the behaviours that has been monitored by the army of so-called ‘big brothers’, state-mandated spies put into Uyghur homes to record ‘extremist’ behaviour, and would likely have resulted in his imprisonment if discovered (Byler 2018). Her discussion of racism against Uyghurs, or her lamenting of the state's closure of minority-language schools, would invite similar consequences for herself.

While the same can perhaps be said for Erkin, the shifting of power relations in which he is enmeshed has taken a strikingly different form. Erkin is one of a small number of my research participants with whom I was able to speak after the crackdown began. Employing the same political dexterity that I had noted, he managed to keep his business open in the initial months of the crackdown by cooperating with the security forces, although the police had begun taking away both customers and staff. Below, I include some brief excerpts from our discussion, to illustrate both how the context has shifted and the complexities of Erkin's case.

At first, he assumed that those being taken were involved in extremist activities or terrorism, but as those he knew well were taken he was disabused of this:

I remember the day very vividly [. . .] – I got a call from the police. [. . .] They were in [my business], and they wanted one of the staff. [. . .] There were two young guys. I knew them; I'd had so many incidents with police I wasn't worried. I invited them into my office. They asked me his name. I said ‘he was a former [staff member] – what's the problem? He's not here’. They asked where he was, and I [told them]. They asked [for] his number. They showed me a form with his ID, his number, all of it. I said yes [that's him], and they left. I didn't call him because his phone would be tapped. But I asked one of the staff to do so, to ask how he was doing. We couldn't reach him. I confirmed he was taken.

I knew this individual. As we later discovered, he had been arrested for sharing on social media a line of poetry, five years earlier, from the Uyghur nationalist poet Abduherim Otkur. Paralleling the issue I'm depicting here, his words from that time – which was the period when I was in Xinjiang – had grown dangerous in the years since, and had come to harm him.

In spring they took another three staff, raiding the business and taking all the computers. Shortly afterwards, Erkin visited his home town:

[It] was much more grim. More cloudy, it felt, and leaving the airport everyone had to stop, everyone had to be dropped off at checkpoints. The worst discrimination I saw at that time . . . they didn't check the Chinese ID, only Huizu [Chinese Muslims]. And I felt very open discrimination. I had never seen that in my life. They didn't do it openly before. And they started doing it openly, and very confidently. Even the chengguan [enforcers below police level, often armed with clubs] were telling people ‘if you're Chinese go, if you're Huizu, stay’. I felt very bad. I couldn't do anything. [. . .t]he arrests in [home town] started earlier, in March. Very well respected, well known, rich men whom I had known personally and who were friends with me – like older brothers – [aged] fifty, sixty. Powerful businessmen – they were arrested overnight. [. . .] One of them was a close relative of one of [a senior regional official]. His family, his son, his wife, everybody [was taken]. And after the visit [. . .], I felt at no point will I be sound and alive without imprisonment.

In the days following, the arrests accelerated:

But the style of arrest was different. In March, they [had come] to me, asked me, then went to arrest him. The second [time], they just went to their homes and arrested them. [In late spring], they arrested two more [. . .] At this time, they'd just call you from the police station and say, ‘you're on the list. Come to the police station without taking anything’. This time, it was lists. You would never know who'd be on the list. And I know those who were taken away never committed a crime. [Here Erkin outlined the details of a number of staff who had been taken]. So the number of staff [who had been arrested] was at least five. I decided that I was leaving. Because if I didn't I would have had to give in my passport and wait for my arrest. And it had been so hard, psychologically, those days. Unbearable. It was unbearable.

And so he fled, with his immediate family. Neighbours in Urumqi sent word that the police had come for him less than a week later. His political contacts and resources were no longer able to protect him, but did allow him to hold on to and use his passport even after most Uyghur passports had been reclaimed by the state. His very successful business was closed down, and he claimed asylum in another country.

The conversation from which the above was taken was a hurried and somewhat unfocused one; Erkin was still distraught. At the time, he encouraged me to speak out about the situation and what I knew, but was unspecific about how his own data should be used. Nor, for a long time afterwards, was I able to clarify: despite his changed circumstances, Erkin had also gone incommunicado.

Consent, To What Ends?

Erkin's life has been transformed by the changes that Xinjiang has experienced in recent years; I must expect that Anargul's life has been transformed just as radically. Yet whereas it would seem prudent to assume, in my ignorance of her actual circumstances, that Anargul finds herself in a more perilous position than when I knew her, Erkin's risk seems to have decreased – if only because his career, his relationships, his business and his ability to ever go home have already been lost. How, under such circumstances, can my interview transcripts and other research data be used in an ethical manner? How can I judge whether it is reasonable to use information from recent conversations on the basis of consent given years ago? Or – to address the issue at a more fundamental level – what can be used to guide such decisions? In this section, I will argue that the ethical logic inherent to the discipline of anthropology – to which I briefly referred above – must be the final guide.

In the context of the variations in willingness and awareness experienced by both of the above participants, it is clear from the outset that the consent that was given was (a) contingent, (b) given in the awareness that there was an uncertain degree of risk and (c) the product of developing relationships between participants and researcher. These relationships were a key factor in the production and modification of consent over the period of my fieldwork. They were necessary, for example, for my participants to be confident both that I was cognisant of and cautious about the risks they were taking, and that I might make reasonable, ethical decisions regarding the factors on which their consent was contingent and the nature of the risks involved. Such confidence – never absolute – is hazarded in the necessary assumption that the researcher will remain in temperament and behaviour more or less as he or she has done through formation of the relationship. It is a product of the participant's assessment of the nature of the researcher and their imagination of his or her future. This is to say that consent is often only formally a question of specifics; in practice, it is as regularly a question of the interpretation of character.

Ethical positions emerging from the relationships that establish such interpretations comprise the core of the ‘situated’, ‘situational’ or ‘localised’ ethics often suggested as an alternative or supplement to formal codes of ethics (see Kovats-Bernat 2002; Verhallen 2016). As I indicated above, my own position during fieldwork was very similar to these. However, while formal processes of ethics administration and such emergent ethics are necessary elements of ethical reflexion in anthropology, they are insufficient – individually or together – to guide decision-making.

Like my interlocutors in Xinjiang – who took (limited) leaps of faith regarding my character and intentions – I would need to take a similar leap of faith if, in using my research materials now, I were to rely on the ‘situated’ ethical positions that developed during fieldwork. Again, it is impossible for me now to contact participants like Anargul and is likely to remain so: some have been incarcerated and she may well be among them. I am not in a position to find out. But while I suspect that she would not consent to talk with me if she knew the conditions that were to come, I cannot say this with any confidence. The context has shifted radically away from that in which I knew her, and any authority such imaginaries might have had has lost its grounding. Perhaps the situation will have hardened her resolve; perhaps, like my second example, Erkin, she has already lost her job, property and extended family and has managed to flee – and, as such, has seen her immediate risk decrease since the time I knew her. Perhaps she is already incarcerated, and I should instead be telling her story explicitly, or demanding her release – as, for example, I have done, along with other scholars – in the case of Rahile Dawut (Scholars at Risk, 2019).

The consent that I had negotiated with Anargul could never have been informed of the context of its potential publication, and provides little guidance. Nor do codes of ethics, such as the current ASA code (2021), give much scope. The imperative to ‘be sensitive to the possible consequences of [my] work and [to] endeavour to guard against predictably harmful effects’ gives no room for the suggestion that materials from before the current crisis should be published, nor that my participant's desire to speak to me in the first place should count for anything. But nor, as I hope I've shown, do the relationships of trust and respect central to situational ethics offer direction.

Ultimately, the inability of either approach to provide a complete model for action is a product of the fact that both are oriented towards ends other than the production of the particular forms of (generally written) knowledge that are exclusive to, and constitutive of, anthropology as a discipline. To use Alisdair MacIntyre's (1984) terminology, such knowledge is an ‘internal’ good of anthropology – neither the discipline nor this product can exist without the other. In contrast, the protection of the reputation of academic institutions or the interests of participants – which formal codes of ethics and situational ethics respectively aim to further – are goods ‘external’ to the discipline, in that they may both be achieved by other means and that they are not inherently necessary to its practice.

It is this internal good, I argue, that provides the ultimate basis for the professional discretion to which, at the last, all formal codes of ethics must defer. This is not to say that the role of the anthropologist is to emerge from the field and then publish and be damned. As MacIntyre argues, the pursuit of goods internal to practices is generative of virtues that serve that good. If our aim is to both produce meaningful, insightful ethnographic material and anthropological theory both now and in the future, then those virtues must include, inter alia, respect, tolerance, kindness and courage – both in fieldwork and in writing. Without such characteristics, access is imperilled, relationships precluded or squandered, and interpretative understanding occluded by sentiment and moral judgement. In this sense, pursuit of that good alone is enough to necessitate treating our participants well.

To this claim it could well be objected that anthropology has been in a continuous process of evolution² since its instantiation as a discipline in its modern form in the early twentieth century, and that it is consequently unhelpfully reductive to speak of any inherent characteristics. Earlier anthropologies – particularly those in the vein of Tylor and other evolutionists who saw themselves as pursuing a ‘reformer's science’ (1871: 410) in an evolutionary paradigm – could well be said to have different goods in mind (and goods not so far from those of contemporary ‘development studies’). However, since the emergence of those norms that have become core to the practice of the discipline – such as extended fieldwork, participant observation, cultural relativism, the pursuit of the emic – the evolution of the discipline has not been away from the internal good I have described but rather in the direction of methodological and theoretical reforms that have moved us closer toward it. This is, in MacIntyre's conception of a practice, to be expected: it is only in the doing of a thing that its internal good can be recognised.

Anthropology's characteristic concern with the specificities of particular places and times, as well as with holistic understandings of communities, makes it somewhat exceptional in this sense, as the kinds of behaviours motivated by the virtues emerging from pursuit of its internal good generally align with the principles of the liberal traditions of the places from which it emerged and in which it is largely found. Formal codes of academic ethics initially emerged in the shadow of inhumane human medical experiments, such as those conducted during the Second World War (Fluehr-Lobban 1994), and these provide a useful point of comparison. If we take the internal good of medical research to be the production of a particular kind of knowledge about the human body that can allow for the development of treatments, then the use of unwilling participants in inhumane experimentation is not unethical in terms of the strict requirements of the discipline. Indeed, such experimentation produced useful data that has long been the source of its own ethical debates (Post 1991). The fact that there is such a clear gap between these disciplinary ethics and the wider ethics of most societies is precisely the reason that a formal code needed to be imposed. In contrast, the inherent ethics of anthropology align closely with the ideals of cultural tolerance and respect that most Western universities seek to enforce, because it is explicitly concerned with the uniqueness of the communities it studies, the denial of access to any of which represents an irrecoverable loss. Conversely, medical research is concerned with the generic and the generalisable, and it is precisely for this reason that it has no inherent care for human life.

Of course, medical researchers may well be horrified by this claim, and make clear that human life is the primary reason for doing what they do. But these are ultimately personal (or ‘pre-theoretical’; Simpson 2016: 123) – rather than professional – ethics, and operate alongside numerous other motivations such as career development, interest and prestige. Personal ethics thus should be thought of as the last (following institutional and situational ethics) of three major influences that may move the practice of anthropology away from the central attractor of its own internal good. Like these others, it is motivated by concerns distinct from those of the discipline itself. I take the paradigmatic example of such influence to be Nancy Scheper-Hughes’ (1995) call for militant anthropology, which she represented as ‘calling the bluff’ of methodological objectivity but ultimately subordinates the ethics of anthropology to the diverse politics of anthropologists.

This is not to say that most anthropologists would make different choices to Professor Scheper-Hughes. By my own lights, intervening to save the life of a whipped thief in a South African township, as she did (Scheper-Hughes 1995), is a courageous and moral action, but I recognise this for what it is – a taking off of the ‘pith helmet’. Similar decisions, in the light of the current oppression in Xinjiang, have been made by many regional specialists, who have entered openly into activism.³ In the last instance, personal ethics – which are both prior and superior to professional ethics – inevitably come first.

The role of the ethics emerging from the practice of the discipline itself, then, must be to provide the ultimate basis for professional discretion – but for this only; all else is beyond their remit. The need for such a circumscribed core of professional guidance is particularly evident in this circumstance, in which the consent given by my informants has clearly gone stale, but remains the case even where the interests of participants is apparent and the applicability of formal codes of ethics obvious, because the anthropologist must still balance the imperatives of both.

To the case in point, then, the appropriate course of action is clear. Consent, or its lack, is strictly relevant in the discipline's own terms only in as much as it serves the cause of the production of the particular form of knowledge constitutive of anthropology. Which is to say, in this case, that the imperative is to anonymise as much as possible, but allow no imaginaries to stand in the way of using the materials co-produced with my interlocutors. Again, personally held ethical positions will not be overruled by such imperatives, but the role of the anthropologist qua anthropologist is clear.

Conclusions

I have argued that efforts to provide alternative approaches to the ethical guidance offered by the much-maligned formal codes of ethics imposed on anthropologists in recent decades fail to provide a complete guide to professional ethics because they are ultimately predicated on the furtherance of goods which are external to the practice of the discipline. A solution is offered by the necessary logic of the practice of anthropology and the virtues that emerge from it. It seems apparent that this is often accepted implicitly by anthropologists in practice: formal codes are inevitably edged around and sometimes ignored, and decisions are made in the field and in writing that run contrary to the wishes of informants. However, making this explicit, I hope, allows it to operate as a touchstone in what is otherwise a confusion of competing claims and interests.

Very shortly before submission of this article, I once again managed to get in contact with Erkin. He was, when asked, quite happy for me to draw on our conversations. In the absence of such a reconnection, however, and in preference to the paralysis of indecision, professional ethics would have offered a solution. Approaches to ethics that take their key from specific fieldwork contexts, or which advocate ‘moral engagement’ rooted in personal politics, are necessary considerations in the determination of how anthropological knowledge should be used in ends other than its own, just as are formal codes of ethics. But while each of these may, pragmatically, come to be the focus of attention through the process of research, writing and publication, we must explicitly recognise that they are peripheral to the good of anthropology.