Counterterrorism or Cultural Genocide? Theory and Normativity in Knowledge Production About China’s ‘Xinjiang Strategy’
‘Wake up! There is something for us to learn here. What they’re saying applies to us, even if it was not meant for us.’
Friedrich Nietzsche, Anti-Education (1872)
Since 2017, the Chinese Communist Party (CCP) has embarked on a rapid and intense securitisation of the Xinjiang Uyghur Autonomous Region. This has included mass extra-legal internment coupled with coercive thought reform, strict surveillance, restrictions on religious life, public struggle sessions, the placement of children in state orphanages, policies for increased Han settlements, limitations on Uyghur births, and much more. What we should call these actions depends on how we think about them, though how we think of them determines what information about them we consider significant. Is it a jobs training programme that went too far? A counterterrorism exercise using indiscriminate target selection and highly intrusive methods? Or is it the destruction of an indigenous Muslim people by a colonial (or totalitarian?) state masquerading as a jobs training programme and counterterrorism exercise?
The recent high-profile paper ‘Counterterrorism and Preventive Repression: China’s Changing Strategy in Xinjiang’ by Sheena Chestnut Greitens, Myunghee Lee, and Emir Yazici, published in International Security (2020) (hereafter referred to as the Xinjiang Strategy paper), provokes exactly these questions, and is an acute demonstration of the centrality of concept formation in our attempts to understand political acts, as well as the unavoidable reflexive and normative dimensions of scholarship when studying how extreme state behaviours affect vulnerable communities.
Greitens et al. put forward a provocative and counterintuitive argument: that the CCP’s own justification for its actions in Xinjiang, as a counterterrorism campaign, offers the most compelling explanation for its changed approach to dealing with the indigenous Uyghur Muslims there. The authors assert that this justification has been mistakenly overlooked as an important explanatory factor by extant scholarship, and that it is so significant that countries wishing to alter China’s behaviour should adopt the language of counterterrorism rather than human rights. International Security is one of the leading journals in the field of international relations and, while the paper does not uncritically reproduce official formulations of the campaign in Xinjiang, it makes a strong case for the utility of these formulations.
The prominence of the paper and the importance of what is happening in Xinjiang naturally invite dialogue. I think the paper offers a useful starting point for exploring fundamental questions in the enterprise of social science—questions of concept formation, theory, and the role of the social scientist in studying potentially genocidal acts against an indigenous population—which have no easy resolution. This paper puts into sharp relief many of these fundamental lines of inquiry, and by examining them we may be enabled to pursue our vocation with greater reflexivity.
The major problematic raised by the paper is: is what we are observing in Xinjiang a counterterrorism campaign using extreme measures, or a cultural genocide cloaked in the gestures of counterterrorism? Put another way: is the CCP’s objective only to prevent further attacks against Chinese civilians by Uyghur militants, or does the Party also seek to end the existence of the Uyghurs as a people? The two goals are not contradictory: if there were no more Uyghurs, there would be no more perceived threat posed by Uyghur militants. Yet we still wish to enquire into what the CCP thinks it is doing. The available data do not allow us to decide whether the CCP’s use of a counterterrorism framing for its repression in Xinjiang is driven by instrumental or non-instrumental motives—yet how we approach that dilemma has empirical, conceptual, and normative implications.
My argument proceeds as follows. First, I recapitulate in more detail the movements in the Xinjiang Strategy paper and discuss the empirical and conceptual issues raised. I then turn to the role of concept formation in political science and the under-theorised conceptualisation of terrorism and counterterrorism in the paper. Finally, I look at what light the critical perspective sheds on this problem, and how it can help to illuminate the relations we are always involved in when selecting one linguistic formulation or one logic of enquiry rather than another.
The Empirics of Inferring Perceptions
The Xinjiang Strategy paper begins with an account of China’s coercive measures against Uyghurs in Xinjiang—but it is one that is primarily limited to those measures that are consistent with the conceptualisation of the campaign as being primarily aimed at counterterrorism. The authors identify their key contribution, i.e. to demonstrate ‘that an overlooked and significant factor that contributed to the CCP’s change in internal security strategy in Xinjiang was its desire to prevent terrorism from diffusing into China via radicalized transnational Uyghur networks’ (Greitens et al. 2020a, 10).
The next two sections respectively summarise the CCP’s repression of Uyghurs in Xinjiang—again only noting acts consistent with counterterrorism—and survey common explanations for those repressive acts, i.e.: increased contention and unrest in Xinjiang, a ramping up of existing assimilationist policies toward minorities, and the rise to leadership of Party chief in Xinjiang Chen Quanguo. They then turn to the favoured explanation, the changing threat of Uyghurs and transnational terrorism, examine the evidence that the CCP expressed concerns over this threat, and on the basis of this evidence argue that this threat has driven the coercive policies against Uyghurs.
Given the central role that CCP perceptions play in supporting this argument, coupled with the inherent vagaries (and perhaps impossibility in any rigorous empirical way) of separating actual CCP perceptions from instrumental CCP propaganda, I argue that more care should have been taken in substantiating their claims about CCP perceptions. 
More importantly, the authors do not use the same rigorous standards of analysis to assess the competing explanations for the CCP’s Xinjiang campaign against the favoured explanation. The main competitor to the counterterrorism theory is the role of Chen Quanguo in instigating the securitisation drive, which began soon after he assumed the role of Party secretary in Xinjiang (Leibold 2019; Zenz and Leibold 2017). The authors write that assessing Chen’s role ‘confronts a fundamental empirical challenge: ultimately, we do not have the information necessary to pinpoint the locus of decisionmaking, the set of policy options considered, the process by which the decision was made, or the underlying motivations of the actors whose preferences were decisive. We cannot know for sure whether Chen is a policy entrepreneur, an implementer of central policy, or something in between’ (Greitens et al. 2020a, 27–28).
Yet this ‘fundamental empirical challenge’ posed by the Chen Quanguo explanation applies just as well to the favoured counterterrorism explanation: in that case, the decisionmaking process and actor motives also inhabit a black box.
We observe a series of violent actions by Uyghurs and overseas jihadist groups, claims by Party-affiliated sources that some portion of rural Uyghurs were vulnerable to extremism, and mass detention justified in part, internally and externally, by reference to these threats. It may seem uncontroversial to draw a straight causal line through these events—but we have no criteria for judging what ultimately accounted for the timing and repertoire of coercive choices. Does mere temporal correspondence suffice? If so, the timing of Chen’s rise and the crackdown would seem just as compelling. We also do not know whether or not the stated justifications functioned as actual reasons—that is, threats perceived by CCP actors who then took countermeasures—or were constructed as threats by Party actors via their own ideological filters and political exigencies. These processes may also be impossible to disaggregate given how theory shapes reality.
The same evidence adduced by the authors could very well account for the security campaign beginning anytime from 2014 onwards, when Xi Jinping visited the region and made the speeches leaked to The New York Times that the authors rely on (Ramzy and Buckley 2019). Xi’s comments are again part of a confusing temporal argument in a follow-up piece by the authors for the Brookings Institution (Greitens et al. 2020b). In it, they refer to contacts between Uyghurs and Islamic militant organisations in Southeast Asia and the Middle East between 2014 and 2016, statements threatening attacks on China by al-Qaeda and ISIS-linked organisations in 2015 and 2016, and note that ‘These developments appear to have gotten the CCP’s attention’. They then cite The New York Times documents quoting Xi that ‘East Turkestan’s terrorists who have received real-war training in Syria and Afghanistan could at any time launch terrorist attacks in Xinjiang’. But The New York Times report indicates that Xi’s speeches were made in April and May 2014. The Southeast Asian contacts between Uyghur and Islamic militants were reported between June and September 2014, and the al-Qaeda/ISIS links one to two years later. Xi could obviously not have been responding to events that took place in the future. Thus, while it is clear that Xi warned about terrorism in 2014, those comments should not be cited as part of a culminating argument, referring to reports months and years after that to show that the timing of the campaign can be primarily explained by ‘internal documents appear[ing] to confirm that senior party leaders, including Xi himself, do fear terrorist threats from abroad destabilizing their rule at home’ (Greitens et al. 2020b).
Ultimately, how we should understand CCP statements of concern about terrorism depends on how we conceptualise the actions they have taken against Uyghurs.
The Problem of Conceptualisation
There are no rules in social science for deciding which type of framework, or set of concepts, an investigation should be couched within. All scholarship operates at the tangled intersection of theory and reality. Theory guides what we see because it provides the categories we use to organise data. Better theory is able to account for a larger number of observations, and thus results in fewer anomalies when fitted to our observations; more extensive fact-gathering and greater scepticism about received frameworks allow us greater theoretical discrimination and more rapid knowledge development (Geddes 2003, 17). But how does a scholar know if their theory is wrong? What stops one from emphasising only the evidence that supports a favoured theory, while discounting other evidence?
Ideally, what prevents this within a positivist research paradigm is rigorous hypothesis testing: exposing a conjecture to falsification based on some explicit criteria. But when attempting to evaluate competing explanations for a complex security policy in a closed authoritarian state, it is difficult to test rival explanations for political outcomes in a rigorous way. Typically the most a scholar can do is put forward explanations that are the most consistent with the facts and which are internally coherent. For the most part the authors have adhered to this template.
The authors’ imagined cost–benefit analysis for the CCP is straightforward: coercively reeducate almost the entire Uyghur population so as to prevent anyone committing terrorist acts, and thus maintain the Communist Party’s power. But no evidence is provided showing the CCP in fact considered the isolated acts of terrorism by Uyghur militants to be a genuine threat to its rule, and this proposed core logic of repression has not yet been thoroughly tested in the literature (Davenport 2007, 9).
I argue that before treating CCP policy in Xinjiang as a case of counterterrorism, the authors should have carefully evaluated what else it may be. An assessment should first establish clear definitions for the concepts of interest, then gather the evidence corresponding to them. We reject the claim that the CCP’s Xinjiang strategy is primarily a jobs training programme because this claim cannot account for so many of the repressive acts. When enough anomalies accumulate, we are forced to adopt a new framework.
But if we imagine two columns, with ‘preventive repression’ being one and ‘cultural genocide’ the other, we would find that many of the items under each of them would be the same. Mass internment and reeducation are posited as the core of the authors’ preventive repression concept; but they could also be a key feature of the elimination of Uyghur identity.
The authors do not mention all of the following, but the Party has: destroyed graves and mosques, forbidden almost all forms of Islamic religious life and expression (failing to discriminate ‘normal’ from ‘extreme’), forbidden the teaching of Uyghur language in schools, systematically denied Uyghurs economic opportunities, expropriated Uyghur lands, encouraged Han settlement, encouraged intermarriage (Han men, Uyghur women), transferred hundreds of thousands of Uyghurs from detention camps to forced labour, and reportedly fostered conditions of mass rape of Uyghur women (Fernando 2019; Hoja 2020; Shir 2019; Stavrou 2019; Tohti 2015; Zenz 2019, 2020). There are reports of local officials forcing Uyghurs to eat pork and drink alcohol during Ramadan (Long 2019). The secular evangelism of assimilationist Han family quartering campaigns since 2014, involving the careful ferreting out of substantially all religious attitudes and customs in Uyghur private family life (Byler 2018), also reveals a level of care and attention for the inner worlds of Uyghurs that is difficult to square with an instrumentalist political goal. Moreover, the special attention given to crushing Uyghur intellectuals mirrors Zygmunt Bauman’s argument that for genocidaires, destroying the traditional elites of the doomed community is key: ‘It is hoped that the marked group, once deprived of leadership and centres of authority, will lose its cohesiveness and the ability to sustain its own identity, and consequently its defensive potential’ (Bauman 1991, 119).
Whether the campaign as counterterrorism fits all of these observations is dubious. The authors do not mention these acts, though a steel manned version of their argument could claim that some of these policies, like ethnic intermarriage and reeducation, can be understood as being within their definition of preventive repression. In this frame, ethnic cleansing and settler colonialism become not genocidal tools but counterinsurgency strategies (McNamee 2018; McNamee and Zhang 2019). Others, like the assaults on religious belief and destruction of mosques, could also be read both ways: as counterterrorism, intended to cut off connections with the wider Sunni world; or as cultural genocide, motivated by the Party’s intent to destroy the Islamic faith of Uyghurs—an anti-religious ideological goal driven by the perceived threat of these beliefs as such (Zenz 2020).
But when enough of these policies are enacted, when should we stop thinking of the CCP’s campaign as a case of extreme counterterrorism, and start thinking of it as cultural genocide?
These problems are archetypal to social science. The philosopher of science Larry Laudan notes that medieval, Renaissance, and seventeenth-century epistemology all shared the core question: ‘To what degree, if at all, does a confirming instance of a theory contribute to the cognitive well-foundedness of that theory?’ (Laudan 1981, 77). One response is to argue that because reality is endlessly narratable, we should adopt narratives that are the most emancipatory. But in that case we seem forced to dispense with any distinction between understanding and action. ‘Take the case of a social scientist committed to a revolutionary movement,’ Peter Berger writes, ‘He may want to shoulder a gun—but others can do this … in terms of his usefulness to the revolutionary movement, the social scientist will be best employed in the task of objective understanding: what is the situation really like? How can its future course be projected … All of these are questions of fact, not of value’ (Berger 1976, 136–37). Berger defines the ideal of value-freeness as perceiving social reality without being influenced by one’s hopes or fears (Berger 1976, 135). This is why we should seek theories that ‘carve nature at its joints’, and insist that concepts in social science not be coterminous.
If we included in the definition of preventive repression the complete cultural destruction, biological assimilation, and psychological conversion of the Uyghur people, simply because it achieves the goal of stopping Uyghur militants, then cultural genocide would simply have been reinterpreted—or defined out of existence.
By assuming an instrumentalist, neo-realist explanation for the CCP’s security actions, the authors limited the evidence they considered to that which could be readily explained by an assumption of instrumentalism. This requires accepting the Party’s own explanation at face value—that its own extreme thought reform programme is indeed ‘deep preventive counterterrorism work’ (Greitens et al. 2020a, 43). It also requires reading against Party discourse and practices that exceed instrumentalist logics, or failing to recognise how counterterrorism rhetoric is subsumed into the Party’s broader dream of (homogeneous) nation-state formation. Absent these considerations, the Party’s attitude toward ‘religious extremism’ becomes merely that it is ‘the ideological foundation of terrorism, implying that the only truly effective form of prevention is altering people’s thinking’ (Greitens et al. 2020a, 43).
The authors’ causal chain as proposed is clear, but the conceptualisation struggles to account for many of the CCP’s actions in its campaign against Uyghurs. The authors concede that ‘China’s threat perceptions may be inaccurate, and/or its public rhetoric may be instrumental’ (Greitens et al. 2020a, 46), yet suggest that ‘those who seek to alter China’s treatment of its Uyghur citizens may be more effective if they approach that behavior as grounded in counterterrorism policy, rather than framing objections on human rights grounds’ (Greitens et al. 2020a, 46).
But what is at question is precisely the validity and legitimacy of the CCP’s threat perceptions. If counterterrorism is indeed being invoked cynically to mask a cultural genocide, or is simply a euphemism for ethnic splittism, the proposed remedies may change considerably. The problem is not simply one of accuracy, but intentionality. Here, questions of normativity intersect with both our means and our ends as students of the political.
Critique, Method, and Meaning
The hypothesis that the CCP saw a practical problem—Uyghurs being infected by extremist religious ideology and committing acts of terrorism because of it—and sought out a practical if radical solution—mass internment and thought reform—would only be valid if we could be confident that the regime would not have undertaken the campaign, or some version of it, in the absence of terrorism. We cannot rerun history without terrorism and see whether the campaign would have been launched, but we can examine past political campaigns in the People’s Republic of China and elsewhere and note the absence of similar motors. The Australian colonialists’ violent destruction and biological absorption of the Indigenous population, for instance, were driven by ideology and interests and did not require a counterterrorism logic (Ellinghaus 2009; Dirk Moses 2004). Nor were the colonisation of Tibet or the reeducation and destruction of Falun Gong—tactics we see combined in the Uyghur case—driven by concerns over terrorism.
There is also nothing deterministic about a violent action against civilians being coded as ‘terrorism’. The act of naming ‘carries the ability to constitute other activities and then codify appropriate actions in response’ (Featherston 2018, 4). The CCP is not a passive, reactive participant in its relationship with its Uyghur minority; the Party shapes the playing field. Party theoreticians are aware of this, as when Li Shulei, director of the Central Commission for Discipline Inspection, instructs that ‘language never only reflects reality; it moulds reality … Language is not a political instrument; it is politics itself’ (cited in Sorace 2017, 10).
The tools of critical theory can be helpful in turning upside down and inside out the problems raised by the CCP’s Xinjiang strategy. To begin with, critique can reveal the contingency of and call into question the coherence of terms like ‘terrorism’, and make claims about who is served by them and in what context; it can reveal how methodological choices are themselves not neutral but instead ‘enact’ and ‘disrupt’ worlds by establishing their own criteria for judging significance (Aradau and Huysmans 2014); it can serve to reframe state acts, showing how it may be that a state less passively ‘perceives’ a threat than it actively constructs one to achieve its ends (hence ‘securitisation’). Most pertinently for us, scholars in critical genocide studies have proposed alternative frameworks to account for seemingly incomprehensible political acts. I highlight two of these.
The first theorises genocide as a social process. Here we are encouraged to understand genocide as ‘a specific technology of power’, a form of ‘social engineering that creates, destroys, or reorganizes relationships within a given society’, and which establishes ‘new social relations and identity models through terror’ (Feierstein 2014, 1, 36). Other scholars have studied how genocidal regimes construct their victims through ideological prisms, reading them as transitioning from marginal insiders to outright foreign, a mortal threat, and subhuman (Hiebert 2017, 137–143). This then justifies eliminationist violence. Scholars have begun applying similar frameworks to the CCP’s construction of Uyghur identity as deviant, with all that entails (Roberts 2018a; Smith Finley 2019). These theories do not account for the timing of the crackdown either, however, and can appear to read back into events a coherent plan to destroy the Uyghurs.
Jacques Sémelin’s notion of ‘delusional rationality’ provides the clearest account of what is at stake when considering the meaning-making of genocidaires. The concept can be summarised in a word—‘Paranoia’. As Sémelin writes: ‘The word fits perfectly, since the structuring of these imaginary representations of the enemy is reminiscent of the descriptions psychiatry ascribes to paranoid personalities. The character traits of a paranoiac can be recognized in a combination of wariness (excessive aggression toward other people), inflexibility (the incapacity to question one’s own value system), a hypertrophied Ego (which can go as far as megalomania), flawed judgment’ (Sémelin 2007, 44–45).
Sémelin is here worth quoting at length:
A paranoiac’s logic is flawed by passion, which leads to a delusional interpretation of reality. Ideas are guided by a preconceived belief. Doubt is as foreign to the paranoiac as self-criticism … The paranoiac reasons correctly on false premises. The starting point seems obvious to him because of a subjective preconception that he is unable to question. Due to this flaw in judgment, the paranoiac always skirts the real problems, seems out of tune with reality and is impervious to other people’s opinions. Yet his intellectual faculties are intact. His discourse does draw on reality, but gives it an imaginary interpretation. The paranoiac attributes all his difficulties and all his failings to someone else. He can’t call himself into question: the world is all wrong, or people are out to get him. But his interpretations are still plausible, and the paranoiac can enlist their support. (Sémelin 2007, 45; emphasis added)
A notion like ‘delusional rationality’ resists the kind of precise measurement that positivist methods insist on, because it is founded in an alternative, interpretivist mode of social inquiry. Such theories illuminate an otherwise obscure puzzle: how can we explain the CCP’s apparent determination to violently erase the Uyghurs as a people, in response to the paramilitary training of a small number of Uyghur militants and terrorist attacks by a smaller number? The militancy itself can also be interpreted as a desperate response to social conditions created by CCP rule (Tohti 2015). In this reading, the crackdown is revealed as a paranoid reaction to a tiny threat, one that is both counterproductive and non-instrumental.
The Politics of Classification
The relevance of these frameworks for the inferential problems I identify in the International Security paper is that they can make us alive to the possibility of non-instrumental, or ideological, drivers of state behaviour. In the language of quantitative social science, failure to consider them would be omitted variable bias. Yet, by definition, they cannot be quantified and modelled, since ideology shapes how actors perceive political reality.
Attempts to shortcut the demand to theorise the normative dimensions of political rationality are likely to end up in the quandary proposed by Marxist thinker Louis Althusser, where ‘what thus seems to take place outside ideology … in reality takes place in ideology. What really takes place in ideology seems therefore to take place outside it’ (Eagleton 2014, 109). Assuming a single model of rationality can only account for the CCP’s Xinjiang strategy when it fails to consider the totality of state action. Reading the campaign as counterterrorism may seem to transform it into a legible problem for policymakers, but we should wish to avoid becoming ‘problem-solvers’ who believe in methods but not worldviews, as Hannah Arendt (1971) put it.
Though social science should be conducted without any specific policy objective in mind, the way social scientists classify political behaviour has political implications. The stance of the Xinjiang Strategy authors appears to be that there is both a social science case and a favourable policy case (i.e. that it will result in less violence and killing of Uyghurs) for treating the campaign as counterterrorism, if only to open a channel for dialogue with the Party.
But it is far from clear that this is true. The pragmatic adoption of CCP classificatory schema, where it conceals the underlying reality, could also just as well perpetuate the status quo and do little to help Uyghurs. The CCP would probably appreciate its own rhetoric about the campaign taking hold and the normative language of human rights atrocities and cultural genocide being abandoned. In that case, counterterrorism conferences and dialogue sessions for Chinese officials may only end up pseudo-solving a misdiagnosed problem. Critical approaches allow us to see past the CCP’s attempts to reframe what is happening in Xinjiang, to flatten the multi-faceted destruction of an indigenous people into a pragmatic if extreme counterterrorism exercise. The precise policy implications of this are up for debate.
I thank Bermond Scoggins, Michael Clarke, Lachlan McNamee, Murray Bessette, Adrian Zenz, and Mustafa Aksu for their comments on drafts of this article. I am grateful to the editors of the Made in China Journal for their attentive editing and feedback. I take responsibility for the arguments and any errors or omissions.
Cover Photo: Barbed Wire Fence, by Pat O’Malley.
 I identify two shortcomings in this regard:
After examining the potential terrorist acts of the groups East Turkestan Islamic Movement and the Turkestan Islamic Party between 2001 and 2014, they write: ‘Thus, until 2014, CCP perceptions and fears of the links among Uyghurs in China, the Uyghur diaspora, and jihadist militant groups remained a theoretical possibility rather than an operational reality.’ This is cited to Raffaello Pantucci in Clarke (2018, 162). But the only reference to CCP perceptions there is the author’s conclusion that al-Qaeda’s 2011 appointment of the leader of the Turkistan Islamic Party as its Pakistan commander ‘did not appear to change China’s impression of threats from al-Qaeda’. This conclusion, in turn, cites an article by the Foundation for the Defense of Democracies (Roggio 2011). That article—the ultimate source of the inference about CCP perceptions circa 2014—was published in 2011 and notes the appointment while making no reference to CCP perceptions. In terms of the authors’ overall argument, this deficit is largely immaterial; it simply shows that we don’t know about CCP perceptions in relation to this information.
Later in the article, the authors refer to ‘CCP thinking in 2017’, when Chen Quanguo outlines six principles for combatting terrorism in Xinjiang. Part of this included reducing contacts between Uyghurs in Xinjiang and jihadists abroad. The authors write: ‘The rise of this approach was based on the CCP’s growing belief that China’s Muslim population was more vulnerable to foreign jihadists than previous assessments had indicated.’ They write that references to ‘infection’ began appearing in 2015 in discussions of Xinjiang, and cite two studies as evidence. But they imprecisely report what these studies say. The first comes from a quote in a Phoenix Television Online article in October 2015 by Zhang Yun, the Party secretary of Xinjiang’s Department of Justice, who said that in southern Xinjiang ‘of those who have been influenced by religious extremism’ (受宗教极端思想影响的人群中), 70 percent have been ‘dragged along’ and 30 percent have been ‘contaminated’ (Chen 2015). These categories appear to refer to the degree to which these populations have been influenced by religious extremism. Zhang refers to ‘a very small number’ of others who have committed crimes. The reference, in any case, was to villages in southern Xinjiang, and it did not state what portion of the total population was claimed to have been influenced by religious extremism (of which 70 percent were mildly influenced, and 30 percent heavily influenced). The authors cited his statement as ‘explain[ing] that approximately 30 percent of Xinjiang’s population [i.e. total population, across the country] had been infected by religious extremism’. A similar oversight is committed in the following reference to a study by the sociologist Li Xiaoxia (2016). Li’s original paper makes a similar taxonomy of victims of extremism, prefaced with a similar formulation: ‘of the communities influenced by “Wahhabism”’ ( 在受“瓦哈比”极端思想影响的群体中). Again we do not know to which subset of the total population these refer. Moreover, Li was referring to the Ili Kazakh Autonomous Prefecture in the north of Xinjiang, and cited a (likely internal Party) document from 2013. The authors write that she ‘estimated the number of people [i.e. of the total population, across Xinjiang] contaminated by extremism at around 20 percent’. Moreover, this source is cited to bolster the argument that the Party’s concerns with ‘infection’ were heightened in 2015. The authors conclude that: ‘These figures track fairly closely with estimates of the percentage of the population detained for re-education.’ Thus, neither of the uses to which these two sources were put quite corresponds with what they said.