“Wash Brains, Cleanse Hearts”: Evidence from Chinese Government Documents about the Nature and Extent of Xinjiang’s Extrajudicial Internment Campaign

Journal of Political Risk, Vol. 7, No. 11, November 2019

Adrian Zenz, Ph.D.
Senior Fellow in China Studies
Victims of Communism Memorial Foundation

Women undergoing “reeducation” in Xinjiang Uyghur Autonomous Region (XUAR, China) were kept in their classroom behind a gated metal fence. See Figure 16, below. Source: anonymous informant.

In the wake of growing international criticism, the Chinese government has sought to counter human rights accusations over its re-education and internment campaign in Xinjiang through an elaborate propaganda campaign. This campaign portrays the region’s network of so-called “Vocational Skills Education Training Centers” (zhiye jineng jiaoyu peixun zhongxin 职业技能教育培训中心) as benign training institutions that offer persons who committed minor offenses an alternative to formal prosecution. Since late 2018, the state has invited media and official representatives from other nations and even from the western media to participate in official and closely-chaperoned tours of a select number of “showcase” centers.[1]

Drawing on the government’s own statements, this article seeks decisively to refute these propaganda claims. Overall, the author analyzed three types of data sources, all of which are mutually consistent and confirm the growing body of first-hand witness accounts. The first type consisted of official government documents and related media reports that are publicly accessible but not designed for international audiences. The second source consists of local government data in the form of detailed tables and spreadsheets that list the fates of thousands of minority individuals. The third source is a confidential, classified Chinese policy document issued by the Xinjiang Uyghur Autonomous Region’s Party Political and Legal Affairs Committee. This highly significant document details how Xinjiang’s so-called “vocational training centers” are supposed to be run.

In combination, these three sources provide us with unprecedented insights into the region’s re-education internment campaign. Together, they decisively refute Beijing’s propaganda claims.

In this article, Xinjiang’s “Vocational Skills Education Training Centers” are referred to as “Vocational Training Internment Camps” (VTICs). This terminology acknowledges that these facilities offer some form of vocational training, although this “training” only constitutes a relatively small part of the whole indoctrination package. At the same time, this terminology clarifies that these extrajudicial facilities function in a prison-like internment fashion.

Specifically, this article will show the following:

  1. According to government statements, VTICs “wash clean the brains” of those interned in them. Those subjected to such coerced brainwashing are referred to as “re-education persons” – the same term used for detained Falun Gong practitioners. The classified document states that detainees who show signs of resistance are to be subjected to “assault-style re-education” efforts.
  2. Specifically, those interned in VTICs are called “detained re-education persons”. Numerous documents make clear that these “trainees” are in involuntary detention. Despite extensive research, the author did not even find a single government document that supports government claims that people willingly consent to being placed into a VTIC, that they sign any kind of agreement to that end, or that they can request leave. The classified document mandates special security measures to ensure that detainees cannot “escape” while relatives visit them in the camps.
  3. VTICs are guarded by large, dedicated police units, including armed police forces. In one case, target employment for security guards was over twice as high as that of the camp’s teaching staff. In another county, the wages of the designated VTIC police force were budgeted to be nearly three times as high as this county’s entire regular vocational education budget. Government regulations and the classified document both specify that VTICs must implement “escape prevention” measures that also apply to prisons. The classified source adds that VTICs must have dedicated police stations, should employ only the “most capable security forces”, and put extremely stringent security and surveillance measures in place.
  4. VTICs are administered by newly established “education and training bureaus” (ETBs) that fall under the authority of the criminal justice system and are funded from domestic security budgets. They are neither funded nor managed by the regular education system. The classified document mandates that every county in Xinjiang must have an ETB. Based on the new data, the author estimates that Xinjiang likely has approximately 1,300 to 1,400 extrajudicial internment facilities.
  5. VTICs represent only one of up to 8 forms of extrajudicial internment in Xinjiang. Detailed local government data sets show that the internment campaign has mostly swept up males, especially household heads. Internment shares in rural Uyghur majority regions (including those sentenced to prison) range between 10 and 30 percent of the adult population. In 2018, the Xinjiang government provided 1.6 billion RMB in VTIC food subsidies to its ethnic minority regions, enough to feed just under one million persons in this particular form of extrajudicial internment alone.
  6. Evidence shows that the internment drive has focused on removing male authority figures from families as part of the state’s coercive social re-engineering campaign. Internment shares of younger women, who often feature in propaganda videos or “model camps,” are typically very low. Consequently, interned populations feature a much higher share of adults aged 40 or higher than those who are not interned.
  7. Overall, the author suggests a new speculative upper limit estimate of 1.8 million or 15.4 percent of adult members of Xinjiang’s Turkic and Hui ethnic minority groups, and a new minimum estimate of 900,000 or 7.7 percent. These figures pertain to all minority adults in previous (since spring 2017) or current extrajudicial internment. While still speculative, the new upper limit is eminently defensible based on existing and new data sets.
  8. Official data proves that the internment campaign has pushed families below the poverty line by depriving them of their primary labor force, and that net population growth rates in southern Xinjiang have dramatically declined since the beginning of the internments.
  9. The classified document specifies that “students” can only “complete their studies” after having spent at least one year in the facilities. Only once they have fulfilled a detailed set of stringent “graduation” criteria do they actually receive an intensive 3-6 months of skills training. This would indicate a minimum term at the VTICs of 15-18 months, which is consistent with the time between the onset of the re-education campaign (April 2017) and the first propaganda videos published on Chinese media channels showing the “successful outcomes” of the camps (October 2018).
  10. According to the classified document, the Xinjiang government considers VTIC work to be “highly sensitive” in nature. All related information is “strictly confidential” and VTIC data material “must not be aggregated” even by its own staff, likely in order to conceal the scale of the internment campaign.
  11. Chinese claims that Xinjiang has no “re-education camps” are simultaneously true and false. They are superficially true in that such denials use a Chinese term for “re-education” that the government itself never employs. However, they are also manifestly false, given there is abundant evidence from government documents that there are several types of dedicated re-education facilities in Xinjiang, and that the officially-stated primary goal of the VTICs is not “vocational training” but “transformation through education”. Government claims that Xinjiang has no “concentration camps” are both semantically and technically false, and contradicted by the state’s own terminology. Even so, the author suggests that the term “re-education camps” (or “internment camps”) is more accurate and helpful.

As China’s internment and related propaganda campaign progresses, this article provides crucial incriminating evidence about the real nature and purpose of the region’s VTIC network. The empirical evidence discussed below should suffice to support significant, concrete actions by the international community against this unprecedented atrocity.

1. Methodology

For this research, the author analyzed data that derives from the extensive reporting requirements imposed by higher government levels on lower administrative levels (county, township, village). This results in local governments producing extensive spreadsheets and tables filled with detailed individual cases. In 2019, the author obtained a cache of over 25,000 files from different government departments. The obtained data pertains mostly to rural minority regions in Hotan, Kashgar and Kizilsu Prefectures, containing data collected or compiled by local government authorities between the years of 2015 and 2019. In order to protect the source, further details on how and where this data was obtained are not being made public at this point. In this paper, the original data references are replaced with source code numbers. All of these references are verifiable and based on original files. The author may choose to make all of this data publically available in the future.

This data often shows real names, ID numbers, ages, addresses and other personal information as mandated by each form (Figures 1 to 4). For example, there are spreadsheets with lists of young persons with their status of study or work, lists of children with both parents in some form of detention and how they are being cared for, lists of couples of mixed ethnicity and whether they still live together, lists of families and their fulfillment of family planning requirements, lists of persons below the poverty line or who are currently (or no longer) receiving minimum welfare payments, or lists of persons who have failed or are unable to repay their government-issued debt.

Some of the files contained in these data sets also give detailed instructions on how to record data about local households. An annotated form (Figure 5) shows local government officials how to create detailed spreadsheet entries for poor households. Information from these types of very detailed spreadsheets is presented in the sections 9 and 10 of this paper.

Figure 1: Example of a spreadsheet containing detailed information on 1,494 persons from no. 12 village in Azatbagh Township (阿扎提巴格乡), Yarkand County, Kashgar Prefecture. The column on the far right indicates individuals’ internment status. The last six digits of the ID numbers were redacted for privacy reasons. Source reference code: 13501.

Figure 2: Frequently, these spreadsheets are displayed on local public billboards. The last digits of the ID numbers were redacted to protect the privacy of the individuals shown. Location: Pilal Township (皮拉勒乡). Source reference code: 13513.

Figure 3: A detailed individual entry with photo, showing the poverty alleviation situation and related poverty alleviation strategies and goals for the household headed by this Uyghur from Zepu County. The last digits of the ID number were redacted for privacy reasons. This document was independently obtained by the author, rather than provided by a source. Source reference code: 13522.

Figure 4: Form detailing the increase in the membership of a poor household through “natural growth” (in the case: birth), from November 28, 2018. Location: Azatbag Township (阿扎提巴格乡), Yarkand (Shache) County. The last numbers of the ID numbers were redacted for privacy reasons. This document was independently obtained by the author, rather than provided by a source. Source reference code: 13525.

Figure 5: 2019 poor households information gathering form. Source reference code: 13602.

These local government files are unique in that it shows that the government’s policies in regards to re-education or other forms of extrajudicial internment do not merely exist on paper, but have been implemented and are creating very real social and economic challenges for the authorities. In also gives highly relevant and incriminating details on the status of individuals or families that can otherwise only be obtained through witness testimonies. However, because these forms are completed by local government authorities and form the basis on which higher level authorities compile regional statistics and reports, the information contained in them is of an official character.

The author’s second data source is a secret, classified government document, dated November 5, 2017, and issued by the Xinjiang Uyghur Autonomous Region’s Party Political and Legal Affairs Committee (zizhiqu dangwei zhengfawei 自治区党委政法委) (Figure 6). This classified document was passed to Uyghurs in exile from a source in Xinjiang, and then directly passed to the author. Its authenticity has been independently verified by several linguists and Xinjiang experts. Based on his experience and through comparison with numerous other government sources, the author likewise considers the language, formatting, and contents of this document to be indicative of its authenticity.

Figure 6: Electronic copy of the original title page of a confidential, internal Xinjiang government document on the nature and management of the VTICs. The reception year and number were redacted from the red stamp in the top right hand corner by the author in order to protect the source. Source: anonymous informant in Xinjiang.

The header of this “telegram” (jiguan fadian机关发电) bears the name of Zhu Hailun, Xinjiang’s Deputy Party Secretary. Issued in 2017, the year when the re-education campaign started, it carries the secrecy level “classified” (miji jimi 密级机密). This level ranks second in China’s three-tier confidentiality classification scheme and denotes documents that, according to the government’s own definition, contain “important national secrets whose divulgence will cause severe harm to the nation’s security and interests”.[2] The document is labeled “new secret transmission no. 5656” and bears the title “Suggestions for further strengthening and standardizing the vocational skills education training center work”. It is also marked as being of the highest urgency level (dengji teti 等级特提) in China’s public document categorization scheme of four levels of urgency, denoting telegrams that must be processed within one day.[3]

The authenticity of this classified document can be further corroborated by comparing its layout and structure to that of other Xinjiang government “telegrams”.[4] The two examples shown in Figures 7 and 8 are not classified, but are likewise labeled as being of the highest urgency and the second highest urgency respectively.

Figure 7: Urgent telegram issued by the Bayingol Mongolian Autonomous Prefecture Party Committee Education Work Committee. This document was independently obtained by the author, rather than provided by a source. Source reference code: 14117.

Figure 8: Urgent telegram issued by the Xinjiang Poverty Alleviation Office and the Industry and Commerce Federation. This document was independently obtained by the author, rather than provided by a source. Source reference code: 14118.

While the contents of this document are consistent with the author’s other findings, its uniquely forthright and candid statements about the true nature of these “centers” provides unprecedented new insights and verification of our previous observations and educated guesses.

Being aware of the discrepancies between its propaganda and the reality, the Chinese state has been using varied and ingenious terms for VTICs in its online publications, some of them evidently designed to obstruct or even prevent targeted keyword searches. For example, some government documents conceal the term “Education Training Center” (jiao pei zhongxin 教培中心), a common short form of the full term, with an asterisk, as in “职业技能*” (zhiye jineng). Other documents replace these four Chinese characters with four ASCII characters, such as “◇◇◇◇”, or else through a mix of Latin and Chinese characters (“JP 中心”) that appear to serve no other logical purpose than obfuscation.[5] The first method is effective for preventing keyword searches, because Google and other search engines cannot actually search for the asterisk character itself.

Generally, Latin abbreviations of incriminating terms appear to have become quite common, especially in the more sensitive local level government files. The document shown in Figure 9 lists legitimate reasons for not participating in the pension insurance system. These include persons who have been “sentenced” [to prison] (PX = pan xing 判刑); who are in “education and training”, which refers to internment in a VTIC or re-education facility (JP = jiao pei 教培); or in detention (SY = shou ya 收押), which typically takes place in a detention center (kanshousuo看守所). Other abbreviations used in the local documents are for example SLRY (san lei ren yuan 三类人员, literally “tree types of persons”, denoting those in re-education, in detention, and sentenced to prison). The only conceivable reason for the use of these Latin abbreviations is the obfuscation of incriminating information.

Figure 9: Hotan County directive about urgently investigating persons who are currently not participating in the government pension scheme. This document was independently obtained by the author, rather than provided by a source. Source reference code: 14206.

Another strategy is to leave incriminating spreadsheet header labels blank when the order of the data columns is well-known. For example, Qapqal County published detailed tables on welfare payments with a breakdown by recipient category.[6] One of the recipient types are “persons in education and training” (jiaoyu peixun renyuan 教育培训人员), a euphemism for persons in re-education or VTICs. The first two months of 2018 show this label. In the tables of the subsequent months, these column headers were simply left blank. Those who are familiar with this data know what data this column contains, but the less initiated will be fooled by this method.

2. Xinjiang’s Vocational Training Internment Camps “Wash Brains”

Official government documentation repeatedly and unambiguously testifies to the fact that Xinjiang’s VTICs engage in known and pre-existing forms of coercive and abusive political re-education. At least five different Xinjiang government or educational institution websites clearly and unambiguously state that VTICs are dedicated brain-washing institutions. To quote:

Vocational Skills Education Training Centers wash clean the brains of people who became bewitched by the extreme religious ideologies of the ‘three forces’.[7]

The 2017 work report of Xinyuan County’s justice bureau puts this in perhaps even more drastic terms. Under the heading “centralized transformation through education work”, the report states that re-education work must “wash brains, cleanse hearts, support the right, remove the evil” (xinao jingxin fuzheng quxie 洗脑净心扶正祛邪).[8]

The method by which the Chinese authorities “wash clean the brains” or “cleanse hearts” is no secret. This Chinese term for the re-education performed in Xinjiang is “transformation through education”, in Chinese jiaoyu zhuanhua (教育转化).[9] The Chinese word zhuanhua, which effectively turns “education” into re-education or indoctrination, literally means to “transform” or to “convert”. In other contexts, it is also used to refer to the chemical process of isomerization by which one molecule is transformed into another.

Transformation through education has its roots in the infamous Re-Education Through Labor (RETL) system which was set up under Mao Zedong in 1957. It is also employed by the Chinese public security authorities to describe the involuntary and often abusive process of coercive isolated detoxification of drug addicts, as well as the torturous treatments administered to Falun Gong practitioners.[10] A 2016 report published by the Xinjiang United Front Work Department describes transformation through education as “drip irrigation” work.[11]

Xinjiang’s governor Shohrat Zakir and others have repeatedly denied accusations that the region is detaining vast numbers of ethnic minorities in “re-education camps”. In these denials, Zakir and others use the Chinese term zai jiaoyu ying (再教育营) that has been commonly employed in Chinese language Western media articles. Such denials are simultaneously straightforward and misleading, because this term is in fact never used in Chinese government documents. Instead, Xinjiang’s “vocational training” initiative is intimately associated with transformation through education and follows related naming conventions.

According to Xinjiang’s “de-extremification” ordinance from March 2017, the primary legal document pertaining to the VTIC network, “de-extremification must do transformation through education well, implementing legal education, thought education, psychological counseling, behavioral correction and skills education, … strengthening the outcome of transformation through education”.[12] Skills education is therefore, even nominally, only one aspect of an evidently highly coercive re-education procedure. Also, despite the fact that this document represents the main legal basis for Xinjiang’s de-extremification work, the term “vocational” (zhiye 职业) never appears in it. In fact, in all relevant government documents and reports published between 2013 and mid-2018, the connection between de-extremification and vocational or skills training is very limited.

It was not until October 2018 that an amendment to Xinjiang’s de-extremification ordinance introduced the concept of “Vocational Skills Education Training Centers” in the context of the region’s de-extremification work.[13] Again, vocational or skills training is only mentioned in connection with the overarching concept of transformation through education. The centers themselves are directly referred to as “re-education institutions” (jiaoyu zhuanhua jigou 教育转化机构), and their stated aim is to “strengthen the effectiveness of transformation through education” (zengqiang jiaoyu zhuanhua shixiao 增强教育转化实效).

The classified document goes one step further by adding that those who have “a vague understanding, negative attitudes or even show resistance” are to be dealt with through an “assault-style transformation through education” (jiaoyu zhuanhua gongjian 教育转化攻坚) to “ensure that results are achieved”.

Recent evidence establishes a close link between Xinjiang’s re-education campaign and the brutal indoctrination methods used against Falun Gong practitioners. This campaign is coordinated by government officials who played a key role in the early re-education campaign against the Falun Gong. They have nearly 20 years of experience in transformation through education efforts that targeted religious believers.[14] These campaigns included the establishment of “legal education classes” (fazhi jiaoyu xuexi ban 法制教育学习班), a concept that is also widely used in Xinjiang, where numerous “legal system training schools” (fazhi peixun xuexiao 法制培训学校) were established as part of the re-education campaign. [15]

An August 2017 document published by the website for combating religious cults of the Urumqi County government describes Falun Gong practitioners as “persons in transformation” (zhuanhua renyuan 转化人员).[16] This exact and other equivalent terms (such as jiaoyu zhuanhua renyuan 教育转化人员, bei shoujiao renyuan 被收教人员, jiaozhuan renyuan 教转人员, or shou jiaoyu peixun renyuan 受教育培训人员) are used in government documents to refer to VTIC “trainees”.[17]

3. Vocational Training Internment Camp “Trainees” Are In Involuntary Internment

The government’s own documents make clear that persons in VTICs are in involuntary internment.

Numerous documents mention persons in custody or detention (shouya 收押) in the same vein as persons in re-education or “training”, often in the context that these two groups’ family members are special needs and receive assistance from the government. For example, a September 2017 report by the Xinjiang Science and Technology Department mentions that the department’s village work team donated rice, noodles, cooking oil and sugar to the “family members of those sentenced and in custody, and trainees” (panxing shouya he peixun renyuan jiating 判刑收押和培训人员家庭).[18] Another report by visiting officials from the Xinjiang Branch of the Academy of Sciences from October 2017 uses similarly explicit language, stating that the visitors had purchased coal for the “relatives of strike-hard detainees and educational training persons” (yanda shouya ji jiaoyu peixun renyuan qinshu 严打收押及教育培训人员亲属).[19] Other, similar reports of village visits even abolish the distinction between those two groups and directly refer to “detained trainees” (shouya peixun renyuan 收押培训人员 or shouya shoujiao renyuan 收押收教人员).[20] The phrases “persons detained in re-education” (bei shou jiao renyuan 被收教人员 or shouya zhuanhua renyuan 收押转化人员or bei shouya jiaoyu zhuanhua renyuan 被收押教育转化人员) are employed in numerous other government documents.[21]

Also notable is that VTIC detainees are classified among the so-called “three types of persons” (sanlei renyuan 三类人员), which include: persons who have been formally sentenced (to prison), persons in detention, and persons in centralized re-education (fuxing renyuan, shouya renyuan, jizhong jiaoyu zhuanhua renyuan 服刑人员、收押人员、集中教育转化人员). [22]

The classified document adds that VTIC “students” are subject to a similar internal management scheme as that implemented by China’s detention facilities (kanshousuo 看守所 or baoliusuo 保留所). The latter classify detainees into three levels based on the severity of their (suspected) crime and their overall behavior, even at times using this classification scheme to place them into completely segregated “management areas” (guanqu 管区): “lenient management areas” (kuan guanqu 宽管区), “regular management areas” (pu quanqu 普管区) and “strict management areas” (yan quanqu 严管区).[23] The classified document does not list the lenient type, instead adding an even more severe category: that of “forceful management areas” (qiang guanqu 强管区). It mandates that “students” must be placed into these categories and related separate areas within the “centers”, with placement based on initial and regular ongoing evaluations of their performance, to be measured by their points (explained further below) and their overall behavior. Each management area is to implement the “education and training” based on different methods and distinct management systems. Evidently, detainees who are considered more problematic are subjected to much harsher treatments and indoctrination procedures. This management system in Xinjiang’s camps is confirmed by former detainees and by the testimony of a former VTIC staff.[24]

The Chinese government claims that VTICs are “just like boarding schools” and that its “students” can regularly ask for home leave.[25] However, the author did not come across a single government document that outlines how VTICs establish voluntary agreements with the “students”, or how and within which parameters the latter can request leave. The classified document also does not mention anything about leave. Instead, it notes that detainees must be allowed to have video calls with relatives at least once a month. However, their ability to connect with relatives is also made contingent on their performance. According to the classified source, this privilege is part of a rewards and punishment system that seeks to motivate detainees to be obedient, study hard, and achieve “genuine transformation”.

4. Vocational Training Internment Camps are Heavily Guarded, Prison-Like Facilities

In Spring 2017, right after the first de-extremification ordinance was issued and people in southern Xinjiang began to be detained by the thousands, numerous counties with high ethnic minority populations in Xinjiang issued construction bids for heavily secured VTICs, with high walls, barbed wire, watchtowers, elaborate internal camera systems, police stations, and even bases for special police units.[26] Some bids blatantly stated that vocational training centers are to function as “transformation through education bases” (jiaoyu jianhua jidi 教育转化基地).

For example, in July 2017, Qaraqash (Moyu) County (Hotan Prefecture) commissioned a large “educational training center” (jiaoyu peixun zhongxin 教育培训中心) that was to include multiple buildings, including a “transformation for education center” (jiaoyu zhuanhua zhongxin 教育转化中心) and a massive 2,074sqm armed police forces facility.[27] Similarly, a district in Urumqi published a construction bid for a 36,000sqm vocational training compound that was to include a surrounding wall, fences, a 500sqm police station, a surveillance and monitoring system, and “equipment for visiting family members”.[28] The latter is a video-based intercom system typically found in prisons. A vocational training center bid for Yengisar (Yingjisha) County (Kashgar Prefecture) was likewise to include a surrounding wall, a surveillance and monitoring system, equipment for visiting family members, and a police station.[29] In April 2018, a district in Urumqi commissioned a vocational skills education training center with a 500sqm police station, fences, and equipment for visiting family members.[30] In some instances, vocational training institutions that may have had an actual educational focus were retrospectively “hardened” for stronger internment capabilities, as with an October 2017 bid for the “centralized closed education and training center” (jizhong fengbi jiaoyu peixun zhongxin 集中封闭教育培训中心) in Nilqa (Nileke) County (Ili Prefecture) to add a security fence and a monitoring and surveillance system.[31]

A Han eyewitness who in April 2017 was involved in construction work near Aksu Prefecture’s Aral Vocational Skills School (akesu shi alaer zhiye jishu xuexiao 阿克苏市阿拉尔职业技术学校), described the security measures of the school. There was a surrounding wall, too high to see inside from the outside, and barbed wire. Stationed within the school grounds was a permanent contingent of about 60 armed police guards with four armored patrol vehicles of the type GA06 used by the People’s Armed Police (Figure 10), along with other police vehicles. The witness had himself been interned in Xinjiang in prior years, and had developed a habit of paying close attention to the security features of internment-style facilities. Until this day, this witness is amazed at the number of armed police and related vehicles stationed at a so-called “vocational training” facility.

Figure 10: These special vehicles used by the People’s Armed Police form part of a regular security contingent that guards the Aral Vocational Skills School. This photo, taken in Beijing, is illustrative. Source: Han eyewitness who lived in Xinjiang for nearly two decades.

According to a document issued by Kashgar Prefecture and to the classified document, all VTICs must by default be equipped with the so-called “five preventative measures” (wufang 五防) “demanded by Chen Quanguo”.[32] One of these is called “escape prevention” (fang tuotao 防脱逃), meaning a facility must be designed so that those inside cannot escape from it.[33] In 2018, two VTICs in Qira (Cele) County (Hotan Prefecture) were budgeted to spend 5 million RMB on “police equipment” (jingyong zhuangbei 警用装备) as part of meeting this “five preventative measures” requirement.[34] The same set of security requirements applies to Xinjiang’s prisons, coerced isolated detoxification centers, and educational correction facilities.[35] Similarly, Sichuan Province mandates the same preventative measures, including the “escape prevention” measure, for its prisons and coerced isolated detoxification facilities.[36] The court of Fuzhou City in Jiangxi Province goes into more detail as to what “escape prevention” means, noting that criminal suspects must be accompanied by two police officers at all times, even when using the toilet.[37]

In the classified document, security measures represent the very first item, and of the “five preventative measures”, “escape prevention” is discussed first. The list of related measures is detailed and long. It includes police stations, police guards, fixed routes for the guards, a one button alarm system, stringent examinations for all entering persons and vehicles. The document also mandates the separation of camps into different sections, which is confirmed by satellite footage that shows that in VTICs and other internment camps, each individual building is typically surrounded by a separate security fence. The doors of individual dorm rooms, hallways, and each floor must be double-locked by two different persons. As soon as doors have been opened to let people through, they must then be immediately locked.

Dorms and teaching rooms must have full coverage surveillance systems with no blind spots. “Students” are strictly forbidden from secretly possessing and using mobile phones, and private exchanges between them and the facility staff are likewise strictly forbidden. VTIC staff must employ information technology in tandem with undercover abilities in order to spy on “students” and prevent them from planning trouble. Camps must have guards on patrol at all times, day and night. Every day, the guards must conduct a risk assessment, investigate potential unforeseen threats, and develop a “joint defense patrol system” (lianfang zhiqin xunluo zhidu 联防执勤巡逻制度) in coordination with all surrounding police stations. VTICs must only employ the “most capable security forces”. Those who do not show a high aptitude for their duties are to be replaced at the earliest opportunity.

The classified document further mandates that special attention must be paid that “students” do not escape during class, meals, showering, using the toilet, seeking medical attention, or when relatives come to visit. When “students” must leave a center for medical or other special reasons, they must be accompanied by guards. Former detainees Mihrigul Tursun and Gulbahar Jelilova both testified that they were accompanied to the hospital by armed guards while in detention.[38]

VTICs are guarded and secured by very large police or security guard units. According to the classified document, dedicated camp police stations and police guards are mandatory measures. In December 2017, Guchung (Qitai) County in Changji Prefecture published a procurement bid for 260 special police unit outfits for its “public security bureau vocational skills education training center security guards” (zhiye jingeng jiaoyu peixun zhongxin bao’anyuan 职业技能教育培训中心保安人员).[39] In November 2017, Artux City recruited 100 assistant police staff for its VTIC (zhiye jingeng jiaoyu peixun zhongxin fujing gongzuo renyuan 职业技能教育培训中心辅警工作人员).[40] Similarly, Poskam (Zepu) County’s education and training bureau budget for 2019 states that the county’s VTICs sought to employ 212 teaching staff, but over twice as many (435) security guards (bao’an yuan 保安员).[41]

Finally, Qira (Cele) County’s 2018 budget shows that of a total of 2,619 assistant police officers (xiejing 协警), 810 are assigned to the county’s VTICs (while the Chinese term does not distinguish singular from plural, it can be assumed that the county has more than one VTIC).[42] Consequently, a stunning 31 percent of the county’s entire assistant police force is tasked with guarding VTIC detainees. That year, VTIC police force wages were expected to amount to 36.1 million RMB, nearly three times the county’s entire spending on regular vocational education.[43]

All of this information corresponds to the witness account of a former Xinjiang police officer who helped to bring 600 handcuffed detainees to a VTIC.[44] What he saw there was a highly secured camp that looked like a prison, with extensive camera surveillance and detainees who were barely recognizable because they had lost so much weight.

Local government data sets further indicate that VTICs employ local Uyghurs as security guards (保安), even as private security guards whose employment is handled by private companies. One Uyghur from Elishbishi Village in Alslan Bagh Township (艾力什贝希村阿尔斯兰巴格乡) was shown to work for the Yakan Security Company (yeerqiang baoan gongsi 叶尔羌保安公司) as a security guard at the Yingbage Village VTIC in Alslan Bagh Township (aersilanbagexiang yingbage jiao pei xuexiao 阿尔斯兰巴格乡英巴格教培学校).[45] In that data set, 24 of 1,703 or 1.4 percent of rural Uyghurs had become VTIC or re-education camp security personnel. While that may on the surface not appear to be a high percentage, it compares to 32 persons who became drivers (of all kinds), or 8 who became hairdressers. Also, if 1.4 percent of rural Uyghurs are needed for guard jobs, then the percentage of Uyghurs who are being guarded in these internment facilities must be substantially higher, likely by a factor of ten or more.

Eyewitness and satellite evidence show how the authorities have even been turning regular vocational schools (hence not VTICs) into internment compounds. A previous section already discussed the security features and armed police detachments at the Aral Vocational Skills Center, a facility whose naming convention indicates a regular vocational training institution rather than a VTIC. In addition, an anonymous Uyghur from Xinjiang, told the author about a late 2018 visit to a facility north of Urumqi, near Beidalu Village (北大路村), that is officially called the “Xinjiang Higher Level Vocational School” (Xinjiang gaoji zhiye xuexiao 新疆高级职业学校).[46] The witness reached this camp by public bus number 3003 and alighted at a bus station with the sign “educational training center” (jiaopei zhongxin 教培中心), the most common term for VTICs.[47]

The same facility was shown in a video taken by the Wall Street Journal in November 2018 during their visit to Urumqi.[48] Both this video and Google Earth satellite footage show a walled compound with watchtowers and large buildings (Figure 11 and 12). Satellite images additionally show large fences around the buildings, a typical feature of the region’s re-education and internment camps. According to the eyewitness, these fences separate the buildings, and their gates are guarded by police officers (Figure 13). Satellite images indicate that the construction of the internment compound and security features around the original vocational education school began in April 2017, and was apparently completed by November 2017.

Figure 11: Satellite footage of the compound, showing watchtowers in the corners, as well as fencing around each building. Date: October 31, 2018. Source: Google Earth.

Figure 12: Screenshot from video footage of the compound from the main road, showing the same buildings, walls and watchtowers as on the Google Earth images. Note how the yellow-black striped wall that separates each side of the road is also visible on the satellite image. Source: Wall Street Journal (https://www.youtube.com/watch?v=6sUEek-u14w).

The eyewitness created a 3D computer model of the compound and of the one building that s/he visited (Figures 13 to 15). The model of the teaching building, labeled “B” in figure 13, shows a classroom where detainees are separated from their teacher by a tall metal fence (Figure 14). From just outside the classroom, the witness personally saw and heard a large group of women, aged about 20 to 70 years, who were learning basic Mandarin Chinese. One of the women appeared to be fragile and elderly, had all white hair and looked very weak. All of the women were Uyghur in appearance, and the informant was stunned when s/he saw that they were caged in like animals. Their male teacher, dressed in a police uniform, looked like a Han Chinese. Right outside the door, a female security guard in police uniform who also appeared to be Han sat on a chair.

Figure 13: 3D model of the compound. Building A is the structure with the classroom shown in figure 14. According to the witness, building B had a sign titled “police officer administrative building” (jingguan bangong lou 警官办公楼). Yet another building, left of building B and visible in the lower left corner of figure 12, is used for interrogations (its basement bearing the label “interrogation room” (xunwen shi 询问室). Source: anonymous informant.

Figure 14: 3D model of the teaching building with classroom, metal fencing, and female security guard outside the classroom door. The witness saw the interior of the classroom through the iron bar doors. Source: anonymous informant.

Figure 15: The eyewitness saw a group of women who were learning Chinese. Source: anonymous informant.

Figure 16: The women were kept in their classroom behind a gated metal fence. Source: anonymous informant.

Outside the compound, the eyewitness met a group of Uyghur women who had come to visit their detained spouses. Because it was around noon, the women had to wait next to the compound until the visiting times resumed in the afternoon. One of the women told the witness that one could only visit the camp every two months or so, and only engage in very short and simple conversations. She stated that during visits, detainees were essentially only allowed to respond to questions with a basic “yes” or “no” response.

While it is impossible to independently verify the interior features of this facility, the external features can be verified from satellite and video footage, and its official name states that it is a “vocational training” facility. Overall, the details of this particular eyewitness account are consistent with numerous other witness accounts and with the general body of evidence gleaned from official documents. For example, the apparent teacher-to-police-guard ratio witnessed by this informant roughly corresponds to the data presented above.

5. VTIC Work is Highly Sensitive and Any Related Information is Strictly Confidential

The classified document features a special section titled “Strictly confidential. The vocational skills education and training center work is highly sensitive in nature.” It states that staff and others are “strictly forbidden to bring video and video equipment such as mobile phones and cameras into the teaching and control places.” Also, any information related to the VTICs, in particular statistical data, must not be “aggregated” or “disseminated”, and are “not open to the public”. It is therefore necessary to strengthen the staff’s “awareness of secrecy” (baomi yishi 保密意识).

The injunction against aggregating data indicates that the Xinjiang government seeks to conceal the scale of the internment campaign. Even VTIC staff must not compile or know this information, which explains why it is so difficult to obtain.

Other documents complain that the work of protecting secrets is not done stringently enough. A file containing a list of “problems” from Hotan County, dated November 1, 2018, notes that government staff have not been stringent enough about “protecting secrets” related to the internment campaign.[49] The document mandates that “no person is under no circumstances permitted to disseminate information about detention or re-education via telephone, smartphone, or the internet”, and officials are “strictly forbidden” to receive “related media interviews” or to engage in any form of “unauthorized disclosure” related to the internment campaign.

Figure 17: Source reference code: 14301.

6. VTICs are Administered by New “Education and Training Bureaus”

In 2018, Xinjiang formalized the re-education system by introducing a new bureau, the “Education and Training Bureau”, abbreviated here as ETB (jiaopeiju 教培局, the full term is zhiye jingeng jiaoyu peixun fuwu guanliju 职业技能教育培训服务管理局).[50] Again, “education and training” is equivalent here to “re-education” (see Appendix). Prior to this, the de-extremification and re-education work had been coordinated by “de-extremification leadership small group offices” (qujiduanhua lingdao xiaozu bangongshi去极端化领导小组办公室).[51]

A widely published propaganda piece cites Kashgar City ETB vice head as explaining the purpose of the local VTIC.[52] Another document from Kashgar Prefecture shows that the deputy head of the prefecture justice bureau (sifaju fujuzhang 司法局副局长) is also the head of the ETB (jiaopeiju juzhang 教培局局长).[53] The 2017 final accounts of the Aksu Prefecture Justice Bureau likewise state that its ETB is part of this institution.[54] Generally, ETBs are listed along with other internal security and law enforcement agencies such as the courts, the inspection bureau, the public security organs, and the justice system.[55] Their budgets are included in domestic security budgets.[56]

ETBs are responsible for overseeing the VTICs and their detainees. One government document about rural medical insurance for example states that the “ETBs must pay for the health insurance of each vocational training and education center student” (jiaopeiju fuze ge jiaopei zhongxin xueyuan can baonafei gongzuo 教培局负责各教培中心学员参保缴费工作).[57]

A government notice issued by Aksu City aptly exemplifies this important distinction between VTICs and regular vocational education (zhiye jiaoyu 职业教育) institutions. It groups the four types of institutions together that are administered by the justice bureau and the public security authorities: coerced isolated detoxification centers (jiedusuo 戒毒所), detention centers (kanshousuo 看守所), prisons (jianyu 监狱) and “Vocational Education Training Centers” (zhiye jiaoyu peixun zhongxin 职业教育培训中心).[58] In contrast, the same document places “Vocational Skills Schools” (zhiye jishu xuexiao 职业技术学校) and other types of schools in a separate group which is administered by the public education bureau. Whereas regular vocational education takes place in the context of the education system, the VTICs are administered by the same entities that oversaw the former RETL system.[59]

7. VTICs Carry Stringent Graduation Criteria and Have a Minimum Internment Period

Eyewitnesses have stated that the VTICs that they were interned in operated a points system based on their behavior and “study” progress. This is confirmed by the classified document, which lists sufficient points as one of several stringent “graduation” criteria.

According to the document, detainees can only “graduate” when:

  • There were few problems during their time of internment.
  • Have spent at least one year in the facility.
  • Belong to the “regular management area” (as opposed to the stricter ones).
  • Have reached the required standard, including a sufficient level of overall points, thought transformation, study achievements, obedience to camp regulations, and other criteria (not specified).
  • Have no other issues or conditions that could impact their ability to “graduate”.
  • Pass a final inspection. This involves a clearance from the Public Security Joint Operations Platform (gongan yitihua pingtai 公安一体化平台), which checks whether “new problems” have appeared in a person’s file. This is followed by a detailed written report, which is evaluated by the ETB and the local VTIC leadership small group. The final approval for “graduation” can then only be given by the higher-level (prefecture or city) ETB.

Importantly, the intensified vocational skills of 3-6 months training only commences upon “graduation”. This provides official evidence for what has been suspected or known all along: that the primary focus of the VTIC work is not on vocational skills training, but on political re-education and linguistic assimilation.

All of these regulations effectively mean that VTIC internment takes place for a minimum period of 15-18 months. Notably, this is consistent with the time between the onset of the re-education campaign (April 2017) and the first propaganda videos published on Chinese media channels showing the “successful outcomes” of the camps (October 2018).

The classified document then states that after their release, the VTIC “students” must be closely followed and surveilled for one year by the local authorities, including the local public security agencies and the justice system. The authorities are to “grasp the true behavior” of those who have been released in a “timely manner” and “manage and control” every single one of them. Evidently, the state is concerned about the future attitudes and actions of the released detainees.

8. VTICs Only Represent One of Multiple Forms of Extrajudicial Internment in Xinjiang

The Chinese government pretends that VTICs are the only or primary type of de-extremification facility in Xinjiang. Officials have repeatedly denied the very existence of “re-education camps”. However, VTICs only represent one of multiple forms of internment and re-education in Xinjiang. Government documents, including public bids, indicate up to 8 different types of facilities listed below. This is corroborated by witness statements. For example, Gulzira Auelkhan said that she spent a total of 437 days in 5 different forms of internment.[60] Her final internment consisted of one week of vocational training, followed by forced labor in a factory. Several other informants interviewed by the Globe and Mail likewise stated that they were kept in an elaborate network of diverse internment facilities, with some being more coercive and brutal than others.

Based on the author’s previous work, we can distinguish several types of extrajudicial internment facilities.[61] In some instances, these facilities might differ only by name and not by function or design. For example, the term “education training centers” is frequently used interchangeably with “vocational skills education training centers”.

  1. Centralized transformation through education training centers (jizhong jiaoyu zhuanhua peixun zhongxin 集中教育转化培训中心)
  2. De-extremification transformation through education bases (qujiduanhua jiaoyu zhuanhua jidi 去极端化教育转化基地)
  3. Transformation through education and correction centers (jiaoyu zhuanhua ji jiaozhi zhongxin 教育转化及矫治中心)
  4. Legal system schools or legal system training schools (fazhi xuexiao 法制学校 or fazhi peixun xuexiao 法制培训学校)
  5. Legal system transformation through education centers (sifaju jiaoyu zhuanhua peixun zhongxin 司法局教育转化培训中心)
  6. Centralized closed education training centers (jiaoyu peixun zhongxin 集中封闭教育培训中心)
  7. Vocational skills education training centers (zhiye jineng jiaoyu peixun zhongxin 职业技能教育培训中心), or educational training centers (jiaoyu peixun zhongxin 教育培训中心), with the latter presumably representing a short form of the former. Referred to in this article as VTICs.

It is possible that these different terms can be roughly grouped into three types of re-education internment: a) transformation through education camps, b) legal system “schools” (camps), and c) vocational training internment camps. However, such classification is only based on the camp terminology as there are no other data points for it.

Notably, the word “centralized” (jizhong 集中), which is found in two of the camp labels on the list, can also be translated as “concentrated”. The Chinese word for “concentration camp” (jizhong ying 集中营) employs this term. The claim that Xinjiang does not run any facilities that can technically be referred to as “concentration camps” is therefore both semantically and factually problematic. People are interned in a concentrated fashion in order to more effectively guard and indoctrinate large numbers of them in a limited amount of space. Concentration is therefore only a means to an end, which is why calling these facilities “concentration camps” is technically correct but conceptually not particularly illuminating. Calling them “re-education camps” has the significant advantage of denoting their ultimate purpose.

An eighth form of internment commonly found throughout Xinjiang are the detention centers (kanshousuo 看守所). These are part of the formal criminal justice system throughout China for the temporary detention of criminal suspects (linshi jiya fanzui xianyi ren changsuo 临时羁押犯罪嫌疑人场所) while the state determines whether to press formal charges. They predate the VTICs. However, in the case of Xinjiang, such neat separations are problematic. Detention centers are legally authorized to perform short-term re-education and educational correction for those convicted of minor offenses, for example in lieu of shorter prison sentences. However, they are not legally authorized to implement longer-term forms of extrajudicial internment. Informants and camp survivors have noted that Xinjiang’s detention centers can in practice keep persons detained for much longer than the 37 days permitted by the law (without formal arrest or a decision to press charges), and are also used as sites of political indoctrination.[62]

Their primary function in Xinjiang appears to be that of determining the status of detained persons. Not surprisingly, some of the most brutal testimonies of physical torture come from detention centers.[63] In 2018, Xinjiang’s detention center capacity was massively expanded. Despite its different legal status and function within the justice system compared to re-education or “training” camps, the detention center network appears to complement the former in the overall large-scale internment and indoctrination effort.

Of particular concern is the fact that at least some facilities that used to be regular vocational training “schools” (xuexiao 学校), and that continue to bear innocuous naming designations, appear to also function as internment facilities. Examples include the vocational schools in Urumqi and Aksu that were mentioned above.

9. Estimating the Scale of Xinjiang’s VTIC and Total Extrajudicial Internment
9.1 Overview of VTIC Structure and Potential Number of Facilities

The classified document mandates that every region (prefecture) and county (or city) must establish a “VTIC leadership small group” (zhiye jineng jiaoyu peixun lingdao xiaozu 职业技能教育培训领导小组), which is to be headed by the county committee party secretary. It must also establish an ETB. This means that every single regional administrative unit at county level and above must establish the administrative infrastructure for supervising VTIC work. This is consistent with the author’s previous finding that most of Xinjiang’s regions, including those where the Han constitute the majority of the population, operate re-education facilities. That fact therefore represents an important finding for estimating the scale of the extrajudicial internment campaign.

Related information for lower levels is difficult to obtain. However, a March 2016 United Front Work Department report cited above states that in Kashgar City, every township and residential district had established a “legal school” (fazhi xuexiao 法制学校), while every village and urban district had set up a “transformation through education room” (jiaoyu zhuanhua shi 教育转化室). Notably, this was almost one whole year before the start of the large-scale re-education campaign. Similarly, a 2014 news report stated that Konashähär (Shufu) County had established a three-tiered “transformation through education base” (jiaoyu zhuanhua jidi 教育转化基地) system at county, township and village levels.[64] Some of these bases were also titled “legal system training schools” (fazhi peixun xuexiao 法制培训学校).

A system of VTICs and/or re-education camps in at least each township and urban district is broadly confirmed by a spreadsheet completed by the Alslan Bagh Township (阿尔斯兰巴格乡) government in Yarkand (Shache) County, Kashgar Prefecture.[65] By listing the current employment status of 1,703 persons, it shows that this and two other neighboring townships, Hoshrap Township (藿什拉甫乡) and Udalik Township (乌达力克乡), have VTICs, along with at least two VTICs in the Yarkand (Shache) County seat: Yarkand Industrial Park  VTIC (shache xian gongye yuanqu jiao pei zhongxin 莎车县工业园区教培中心) and Yarkand County Southern City District VTIC (shache xian chengnan jiao pei zhongxin 莎车县城南教培中心). Moreover, Alslan Bagh Township has both a VTIC and a “re-education center” (jiaozhuan zhongxin 教转中心).

This implies that at least in regions with high ethnic minority population shares, the re-education system is very fine-grained. But even in Qapqal (Chabuchaer) County, where Turkic minorities make up only slightly more than 50 percent of the population, a letter to the county government complains about the fact that the county’s townships are suffering from a serious lack of cadres because they have been deployed to the VTICs.[66]

Xinjiang has a total of 119 city-level, prefectural and county-level administrative units, along with 1,079 township-level administrative units. If each county and township had just one re-education center or VTIC, the region’s camp network would number around 1,200 facilities. Even if many townships with Han majority populations did not have such camps, we know from the data presented above and from witness accounts that ethnic minority regions often have several such facilities, especially at the county or prefecture center levels. The new information that has transpired since the author’s first research paper on the re-education campaign (May 2018) therefore indicates that the number of Xinjiang’s re-education facilities is likely close to the original estimate of 1,200. Such a higher end estimate is even more likely given that at least two institutions that are labeled as regular vocational education schools, the Xinjiang Higher Vocational School in Urumqi, and the Aral Vocational Skills Center, are said by witnesses to function as internment camps.

In addition, Xinjiang has at least 119 detention centers, one per administrative unit above township level. Likely, there are more than that. That means that the region has probably somewhere between 1,300 and 1,400 extrajudicial internment facilities (excluding prisons).

9.2 Increases in Local Internment Shares

9.2.1 Internment Shares 2015-2016

New documentary evidence indicates that the total number of persons in Xinjiang who have gone through some form of internment since 2016 or 2017 must indeed be very substantial. First, data from the years before the main re-education campaign (2015 to 2016) are presented. Data from the time during the campaign (2017 to 2018) are discussed subsequently.

A government notice from Akdöngjemi Village (阿克墩结米村) in Sëriqbuya Township (色力布亚镇), Maralbexi (Bachu) County about political and social control work in villages mentions the use of the 4th re-education facility on the list: “strengthen re-education, send focus persons and daily management persons in separate batches to the township legal system training school (fazhi peixun xuexiao 法制培训学校) to implement transformation through education.”[67] The context makes it clear that this form of re-education targets persons that are considered to be more “problematic”. The document also indicates the scale of the detentions. In that particular village in 2016, out of a total population of 1,750, 110 were subjected to re-education in the legal system training school. This figure represents 6.3 percent of the total population or 9.3 percent of the adult population.[68] Notably, this was before the onset of the massive re-education internment campaign in spring 2017.

A government report from Tekes (Tekesi) County, mostly inhabited by Kazakhs, states that by late 2015, the county had 722 “re-education focus persons” (zhuanhua zhongdian renyuan 转化重点人员).[69] That year, the county numbered 20,010 Uyghurs; Kazakhs and other Turkic or predominantly Muslim ethnic groups were not yet targeted. Nearly all of the 722 “re-education focus persons” would have been Uyghur, equivalent to about 5 percent of the adult Uyghur population. Again, this was over one year before Chen Quanguo’s re-education campaign started.

9.2.2 Internment Shares 2017-2019

Similar data from 2017 and 2018 indicates that internment shares have increased significantly since then. In a number of Uyghur majority population regions, they have risen dramatically.

Local spreadsheets from six villages in three townships located in Yarkand (Shache) County detail the situation of 1,138 households with 5,407 persons in 2018, all of them ethnic Uyghurs.[70] The youngest recorded person was born in November 2018. The lists do not specifically target persons in poverty or special needs categories. However, they pertain to Uyghur villagers, a population group that has been a prime focus of the re-education campaign.

Of the 3,249 adults, 511 or 15.7 percent were classified as “three types of persons” (san lei renyuan 三类人员), which therefore includes those sentenced to prison. Internment shares vary greatly by types of persons. Within this data set, 25.4 percent of household heads and 17.3 percent of sons or sons-in-law were interned, compared to only 4.1 percent of spouses (wives) and 2.4 percent of daughters or daughters-in-law.

Contrary to Beijing’s claims, the re-education and internment campaign clearly targets the authority figures as opposed to just the younger generation who are supposedly in need of “training”. The oldest person categorized as “three types” was 77 years old, and the sample included 112 persons aged 50 years or older in this category. A histogram of the data set shows that between the ages 25 and 50, each 5-year cohort is almost equally strongly represented. Only the 5-year population cohorts under 25 and over 50 are underrepresented. Remarkably, the sample includes slightly more interned persons in the 50-55 year cohort compared to the 20-25 year cohort. By comparison, the age cohort distribution of those not interned is strongly skewed towards younger age cohorts.

Figure 18: Source reference code: 14402.

This drastic difference is also reflected in other data sets. Figure 18 shows age cohort breakdowns from three data sets: a) a sample of 3,264 persons that were reported as interned in a list compiled by the Norwegian Uyghur Committee in June 2019; b) the Xinjiang Victims database shahit.biz maintained by Gene Bunin, and c) the large government document sample from six villages discussed above.

Among the three data sets showing persons in internment, the share of those aged 40 years or more ranged between 49 to 57 percent. By comparison for the data set showing persons not in internment, that same share stood at only 34.5 percent. In the internment data sets, the shares of those aged 40 and higher also exceed that of the 2010 census (of 46.0 percent) of the general population in countries with a significant Turkic minority population share.

Data set and sample size Share of those aged 40 and above
Interned (Norwegian Uyghur Committee data set, 807 entries) 57.0%
Interned (Shahit / Xinjiang Victims Database, 865 entries) 49.9%
Interned (local data sets from Yarkand, 511 entries) 49.0%
Non-interned (local data sets from Yarkand, 2,819 entries) 34.5%
2010 census (counties with high Turkic minority population shares, 9.82 million entries) 46.0%

Table 1: Shares of adults aged 40 and above. Sources: Norwegian Uyghur Committee; Xinjiang Victims Database; local government spreadsheets (with source reference codes).

Further clarity on the age distribution of those in internment or prison is achieved by comparing the average age cohort shares across the three data sets of interned persons with those from the 2010 census (calculated based on counties and regions with a high Turkic minority share). The three data sets were combined by calculating a non-weighted average (in order to prevent the much larger data sets from the Norwegian Uyghur Committee and from the Yarkand sample from skewing the average). Figure 19 shows that those in internment have a lower representation among those aged 20 to 29 years, and also among those aged 65 and higher. In contrast, those aged 30 to 54, and especially those aged 45 to 54, are heavily over-represented. The discrepancy between the two data sets is especially marked for the group of those aged 20 to 24 years, which is supposedly the target of Xinjiang’s “vocational training” campaign. This indicates that the strategy of interning (and imprisoning) the middle-aged population, notably household heads, is a region-wide strategy.

Figure 19: The averaged interned age cohort shares were calculated by averaging shares of the three interned persons data sets shown in Table 1 (non-weighted).

In Beijing’s propaganda camps that are showcased on state TV and to foreign visitors, young women make up a prominent share of those in “vocational training”. However, according to this data, they represent the population group with by far the lowest internment shares. Overall, the internment share of all male adults in the local government document sample was 9.2 times higher than that of female adults (90.2 versus 9.8 percent). In the data sets compiled by the Norwegian Uyghur Committee and Gene Bunin, males made up 72.2 and 77.3 percent of those interned. Consequently, Beijing’s claims that the internment campaign primarily aims to provide vocational skills for young adults can be decisively refuted as false.

The local government data also shows that there can be considerable variation even between townships of the same Uyghur majority population county. The spreadsheet from no. 12 village in Azatbag Township (阿扎提巴格乡), Yarkand (Shache) County, showed an overall adult internment share of 28.4 percent, nearly twice as high as that of the entire sample. At the same time, this sub-sample confirms the overarching internment strategy of primarily detaining household heads and other male household members. The internment share of household heads was again by far the highest with 50.5 percent, followed by “sons” (20.0 percent), wives (4.2 percent) and “daughters” (3.1 percent).

Total adult internment shares in the six villages of this data set ranged between 8.5 and 28.4 percent. While the weighted average internment share of this sample stands at 15.7 percent, the non-weighted average across the six villages was 17.5 percent. The non-weighted average represents a relevant alternative evaluation of total internment, because village size and internment shares are not logically correlated. In a weighted average, higher internment shares in smaller villages are underrepresented. In this sample, the village with the lowest internment share could be somewhat of an outlier, because its internment share was significantly lower than any other village, but by virtue of having the largest population, it skewed the total sample average towards the lower end. If we remove this village from the sample, the resulting sub-sample of the five remaining villages had a combined (weighted) internment share of 18.2 percent. This consideration is underscored by the fact that the other internment data sets contained in the document cache featured adult internment shares between 16.9 and 21.3 percent.

Samples from completely different Uyghur (and Kazakh) minority regions show a similar pattern and likewise high internment shares. Household data from Karasu Village (喀拉苏村) in Pilal Township (皮拉勒乡) in Akto (Aketao) County, Kizilsu Kyrgyz Autonomous Prefecture, dated 2018, show that 27.6 percent of household heads, 25.0 percent of “sons”, 4.7 percent of “daughters” and 2.9 percent of spouses (wives) were interned or sentenced (“three types of persons”).[71] Overall, 16.9 percent of the 586 adults in the sample were listed as “three types of persons”. Again, those interned are more heavily represented in the age cohorts 30 to 54 years, while the internment probability of individuals in the 20 to 29 year age cohort are lower (Figure 20).

Figure 20: Source reference code: 18761.

Other data sets do not break down internment shares by person. For example, a spreadsheet from Ayköl Town (阿依库勒镇) in Aksu City from August 28, 2017 shows the reasons why poor households no longer receive minimum welfare payments.[72] In 21.3 percent of all cases (126 of 592 persons, only counting adult family members), the stated reason was internment or prison. Similarly, in Oyluk Village (乌依鲁克村), Yantak Township, (央塔克乡), Makit (Maigaiti) County (Kashgar Prefecture), “three types of persons” households made up 91 of 456 or 20 percent of all households in 2018.[73]

Overall, it is clear that these data sets pertain to rural Uyghur majority population regions that represent the core focus of the internment drive. From this, it appears that broadly speaking, 10 to 30 percent of the rural Uyghur adult population are typically in some form of extrajudicial internment or formally sentenced to prison. In urban areas and among Uyghur populations in regions with higher Han population shares, internment rates are likely lower. However, there is no data to substantiate this.

9.3 Internment by Type

One document shows the internment shares for villages that were implicated in Yarkand’s July 28, 2014 attack, details of which have been heavily censored by the authorities. A report about “issues with poverty alleviation” from October 2018 names two locations in that county, Elishku Town (艾力西湖镇诺其巴扎) and Akbash Village (阿克巴什村), both in Kosherik Township (阔什艾日克乡), that are “heavily polluted by extremist ideology” and were involved in that incident. A stunning 59.0 percent (440 of 746) of all households in these two villages are “three types of persons households”. Only the figures for Akbash Village shows a breakdown by type of internment, indicating an unusually high imprisonment share. Of the 278 “three types of persons households” in that village, 110 had persons sentenced to prison.[74]

In this unique case, the share of “three types of persons” in prison is unusually and unrepresentatively high. All other data sets reviewed by the author indicate that the vast majority of them are in re-education. Also, there are typically many more persons in extrajudicial detention (detention centers) than in prison.

Data set and location Re-education (收教) Detained (收押) Sentenced or in prison (判刑, 监狱)
Ayköl Town (阿依库勒镇), Aksu City (source reference code: 14791)


106 of 126 (84.1%)
17.9% of sample population
14 of 126 (11.1%)
2.4% of sample population
6 of 126 (4.8%)
1.0% of sample population
Ish’hala Town (伊西哈拉镇), Kuqa (Kuche) County (source reference code: 14808) 93 of 97 (95.9%)
9.4% of sample population
0.4% of sample population none
Yengiarpa Village (英阿尔帕村), Pilal Township (皮拉勒乡), Akto County 39 of 40 (97.5%)

15.6% of sample population

1 of 40 (2.5%)

0.4% of sample population

Akbash Village (阿克巴什村), Kosherik Township (阔什艾日克乡), Yarkand County (source reference code: 15116) 143 of 278 (51.4%)

32.3% of sample households

25 of 278 (9.0%)

5.6% of sample households

110 of 278 (39.6%)

24.8% of sample households

Table 2: Breakdown of interned persons by “three types”.

9.4 Government Subsidies as Proxy Indicators of Internment

An important proxy indicator of internment are government subsidies that target those interned or their affected family members.

For example, Qapqal (Chabuchaer) County’s welfare figures show that minimum living standard guarantee (dibao 低保) payments rose 11.3 percent between January and July 2018.[75] The related tables also show dibao payments to “education and training persons (jiaoyu peixun renyuan 教育培训人员)”, a euphemism for VTIC detainees. More accurately, this would be dibao payments to the relatives of detainees, as is evident from many local government tables and spreadsheets. These payments rose by 220.6 percent during that time period, making up 36.7 percent of all of July’s dibao payments. In some townships, this share even exceeded 60 percent, likely because they are home to more Turkic minorities. These minorities’ share of Qapqal’s population is only slightly over 50 percent. Such dramatic increases in dibao payments may be due to a policy change, or to increased internment, or both. In either case, they again indicate a high internment share of poorer rural minority populations.

A very important data point in regards to the scale of Xinjiang’s internment campaign comes from its regional budget. This document shows that in 2018, the regional government gave 1.59 billion RMB of food allowance subsidies for VTIC persons (zhiye jiaoyu peixun xueyuan 职业教育培训中心学员) to its ethnic minority prefectures, most of it to Uyghur majority regions.[76] This pertains only to VTICs and not to other forms of extrajudicial internment. It is possible that the prefectures and counties supplement this further with their own funding, although the author did not analyze this in detail. Witness accounts note that VTIC meals are extremely low in quality and quantity. [77]  If we take the Chinese military’s food allowance for ordinary soldiers of 11 RMB per person per day as the standard and assume that local regions do not furnish any additional funds, these subsidies could have fed about 395,000 persons.[78] If the cost (and quality) of the food given to VTIC detainees is only half that of the average PLA soldier, which appears more than likely, then that figure doubles. If detainees were on a very poor diet below common calorie intake requirements for adults, and we assumed a daily food allowance of 4.5 RMB (such as 1.5 RMB per meal with three daily meals), the VTIC detainee figure alone could be just below one million. Given that the VTICs appear to be one of the most prevalent forms of internment, such numbers are generally within a realistic range.

Specifically, Kashgar Prefecture alone received 666 million RMB in such food subsidies, sufficient for about 166,000 VTIC persons at the average PLA soldier’s daily allowance level, which equates to approximately 5.4 percent of that region’s adult population (in 2018).[79] Again, these estimates refer only to VTICs, and therefore to only one of up to 8 forms of extrajudicial internment. If the daily allowance were less than at the PLA, which is more than likely given witness statements, or if the prefecture added its own funding to this budget, then the number of VTIC detainees alone could be around 10 percent. Together with other forms of internment, Kashgar Prefecture’s internment share could be significantly higher than 10 percent. The same figures for Hotan Prefecture are very similar. With VTIC food subsidies of 360 million RMB and assuming the daily PLA food allowance rate, the share of its adult population detained in VTICs alone would amount to 5.5 percent (about 90,000 persons). Again, the actual share is likely substantially higher than this rather conservative estimate.

In January 2018, Radio Free Asia cited a Kashgar security chief as saying that that region had about 120,000 persons in internment.[80] Given the dramatic expansion of different internment facilities throughout 2018 and the fact that this figure may only have pertained to a subset of all the different forms of internment, the internment number by the end of 2018 could have been substantially higher than 120,000.

Virtually every Uyghur family has at least one, and in numerous instances several family members in detention. Visiting reporters and academics have reported observing empty streets, deserted bazaars, boarded-up homes and shops, and many fewer men than women on public streets.[81] Detentions among Kazakhs and Kyrgyz have also drastically increased, especially in 2018. Large numbers of all of these ethnic groups are kept in the detention centers mentioned above, whose total capacity has drastically risen in the past two years.[82] Witnesses report that all or many of these facilities are severely over-crowded. Spending on Xinjiang’s justice system facilities and detention center management and related domestic security budgets multiplied in 2017.[83]

9.5 Calculating Updated Extrajudicial Internment Estimates

Together with the new data presented in this article, the author considers it necessary to increase the estimate of those who are and have been directly affected by Xinjiang’s extrajudicial internment and re-education drive from the previous speculative upper limit of 1.06 million.[84]

The local government spreadsheets presented above indicate that regions with substantial shares of Turkic minority populations can have combined adult internment and imprisonment shares of about 15 to 20 percent or more. Of these, only about 5 percent are typically sentenced to prison, with about 95 percent being in extrajudicial “vocational training”, transformation through education or in detention centers. In regions where Turkic minorities represent only a small share of the population, these shares may be reduced.

The author’s May 2018 upper limit estimate of 1.06 million persons interned was based on a 10 percent internment share of Uyghur and Kazakh adults aged 20 to 79 years, using data from the 2010 census, and adjusted for population growth between 2010 and 2015 based on updated population figures from the 2015 mini census for Xinjiang.[85] Based on a leaked document published by the Japanese Newsweek and other information obtained by Radio Free Asia, a 10 percent internment share was assumed for regions with a Uyghur and Kazakh population share over 50 percent, and a 5 percent internment share for regions where this share was below 50 percent.

The information that has become available since then, including the new data presented in this paper, indicate not only that internment shares in regions with a high Turkic minority population range anywhere between 10 to nearly 30 percent; numerous witness accounts indicate that internment shares in regions where Turkic minorities constitute only a small share of the population, such as Karamay or Urumqi, have risen significantly since 2017. It is no longer safe to assume that internment shares there are drastically lower than in Uyghur or Kazakh majority population areas. As stated above, the classified document mandates the establishment of VTIC administrative structures in every county in Xinjiang. Importantly, regions with high Han population shares have also seen a substantial increase in detention center capacity. Witness accounts indicate that these types of facilities are used to intern large and growing numbers of ethnic minorities for re-education purposes. Xinjiang’s domestic security spending on detention center management grew by 239 percent between 2016 and 2017, with the increase in a sample of Turkic minority regions amounting to 304 percent.[86]

The growth of both re-education and detention facilities between late 2017 / early 2018 and early 2019 can be demonstrated using satellite imagery. Below, three pertinent examples are presented.

The first example is the Qaghiliq (Yecheng) County Legal System Transformation Through Education and Vocational Training School. The construction bid issued in August 2017 called for the construction of eight buildings with a combined total of 82,000sqm floor space (top left section of 21), next to the existing detention center (bottom left section). Between March and October 2018, the middle section between the re-education camp and the detention center saw the addition of several buildings (red rectangle in the middle of the right section). Notably, the capacity of the detention center was approximately tripled, with the center size increasing from four to 12 buildings.

Figure 21: Qaghiliq (Yecheng) County Legal System Transformation Through Education and Vocational Training School. Source: Google Earth, 37°55’00.40″N, 77°21’05.30″E.

The second example is the Yengishahar (Shule) County Legal System Transformation Through Education School (fazhi jiaoyu zhuanhua xuexiao 法制教育转化学校) in Baren Township (巴仁乡), village no.13. The satellite footage from April 2018 shows five identical buildings (Figure 22; construction bids for three of these buildings were cited in the author’s September 2018 publication).

By May 2019, the overall floor space of the entire compound had increased by 535 percent, from 24,715sqm to approx. 156,800sqm, not counting some of the smaller auxiliary buildings or the blue-roofed buildings that are likely factories. Seven large buildings, about twice as large as previous ones, were added on the west side of the existing ones (along with a large number of blue factory buildings), and four equally sized buildings to the northeast (Figure 23). These large buildings are about two-thirds as long as the United States’ infamous Alcatraz Federal Penitentiary (Figures 23 and 24). To the south, nine smaller almost square buildings were constructed, along with several additional smaller buildings.

Figure 22: Yengishahar (Shule) County Legal System Transformation Through Education School (fazhi jiaoyu zhuanhua xuexiao 法制教育转化学校), April 2018. Source: Google Earth, 39°21’28.05″N 76° 3’6.85″E

Figure 23: Yengishahar (Shule) County Legal System Transformation Through Education School (fazhi jiaoyu zhuanhua xuexiao 法制教育转化学校), May 2019. Source: Google Earth, 39°21’28.05″N 76° 3’6.85″E.

Figure 24: Comparing the new buildings at the Shule County Legal System Transformation Through Education School with the Alcatraz Federal Penitentiary (both at the same scale). Sources: See Figure 23.

The third and final example is the Hotan Legal System Education and Training Center (hetian xian sifa ju jiaoyu peixun zhongxin 和田县司法局教育培训中心). This facility was identified by Shawn Zhang as “camp number 5”, based on public bid documents, and visited by the New York Times in the middle of 2018. [87] Between November 2017 and June 2019, the entire compound not only saw the addition of what appears to be a second re-education camp or VTIC (Figure 26, left side), but also an enlargement of the detention center from three to nine main buildings (Figure 26, right side). Moreover, numerous other building sites were added, including what might be factories (blue-roofed buildings) and another type of training facility (featuring a large sports field). The resulting total compound and floor space areas are multiple times larger than the entire Alcatraz Island (Figure 27). The new detention center (on the far right) by itself is several times larger than the Alcatraz prison.

Figure 25: Hotan Legal System Education and Training Center (hetian xian sifa ju jiaoyu peixun zhongxin 和田县司法局教育培训中心), November 2017. Source: Google Earth, 37°15’1.93″N 79°50’49.13″E.

Figure 26: Hotan Legal System Education and Training Center (hetian xian sifa ju jiaoyu peixun zhongxin 和田县司法局教育培训中心), June 2019. Source: Google Earth, 37°15’1.93″N 79°50’49.13″E.

Figure 27: Comparing the Hotan Legal System Education and Training Center (June 2019) with the size of Alcatraz island and its Federal Penitentiary.

Consequently, the author’s new upper limit extrajudicial internment estimate (including detention centers, excluding prisons) is based on the following assumptions: internment shares in regions with a Turkic (Uyghur, Kazakh, Kyrgyz) minority population of 40 percent or higher are assumed to reach an average adult population current extrajudicial internment of 16 percent; those with a Turkic minority population share between 30 and 40 percent have an assumed internment share of 12.5 percent; and those below 30 percent an assumed internment share of 10.0 percent. Moreover, the calculation assumes a generic internment share among the Hui, a Muslim minority group, of 4.5 percent. Despite the fact that the Hui have a long tradition of assimilation into Chinese language and culture, there are growing indications that the Hui are also being targeted.[88]

These internment share estimates are then multiplied by adult population figures from the 2015 mini census. The results are again multiplied by adult population shares derived from the 2010 census (by county), given that the 2010 census is the most recent reliable data point for population breakdown by age cohort.[89] The total internment estimate figure is then adjusted for the XUAR’s population increase from 23.22 million in 2015 to 25.3 million in 2019.[90] The 2019 estimate was calculated based on a 2018 population of 24.87 million and an average reported annual population growth of 1.75 percent between 2015 and 2018. As indicated in section 10.1, Turkic minority population growth has apparently slowed down dramatically in 2018 (and certainly 2019). However, the growth rates used for calculating internment estimates do not rely on total population increase through new births, but rather on the natural increase of the adult population as larger younger age cohorts naturally reach higher ages.

These calculations result in an estimated concurrent upper limit extrajudicial internment figure among Xinjiang’s Turkic and Hui minority adult population for 2019 of 1.6 million persons.

In order to arrive at an estimate of all persons who are currently or have previously been in such internment, we need to estimate how many persons may have been released from internment between 2017 and 2019. This includes all forms of “release”, including forced labor placements. Based on the internal document “Karakax List”, which indicates that significant shares of detainees had been released by the middle of 2019, a very conservative release share of 10 percent persons is used, with the assumption that most of those would have been “released” into forced labor placements.[90b] As time progresses, it is likely that the numbers of those released are growing, while the numbers of those in current internment are decreasing. This yields an estimate of 177,000 “released” persons.

Adding 177,000 to the current internment estimate of 1.6 million results in a combined figure of 1.777 million, or approximately 1.8 million. This means that 15.4 percent of the adult Turkic and Hui minority population are or have been interned. This is equivalent to just below one in six members of that population, with the difference to the author’s previous estimate from July 2019 of 1.5 million being explained by using updated population figures, including the Hui population in the sample, and basing internment share estimates on the new data sets presented above.

An unknown subset of these persons will have been interned in the VTICs alone. Claims by the Chinese government that the VTICs alone do not contain one million persons are therefore likely but highly misleading. They constitute no refutation of the estimation that well over one million persons are or have been interned in all extrajudicial internment facilities combined.

This new speculative upper limit internment estimate can be corroborated by considering the total number and possible average capacity of Xinjiang’s extrajudicial internment facilities. As discussed above, we must assume that the region has likely up to 1,300 or 1,400 extrajudicial internment facilities of all types. At least several of these have seen a massive enlargement of their capacity, and numerous witnesses testify to the fact that they are incredibly overcrowded.

If Xinjiang has or had up to 1.6 million persons in extrajudicial internment at any given time, that would mean that on average, a facility would intern approximately 1,200 persons. Due to the extremely crowded conditions, this is commensurate with a facility that was originally built to intern 300 to 400 persons but is now interning three to four times that number. Also, at least some re-education camps and detention centers are much larger than that and have several thousand persons interned. If we assumed that the 200 largest extrajudicial internment facilities hold an average of 4,500 persons each, then the remaining, smaller facilities, many of which would likely be located at lower administrative levels, would only have to intern about 600 persons, and may in fact have been designed to only intern about 200 persons under non-crowded conditions.

That is still a fairly conservative estimate, given that the four or five floor buildings of a typical size for re-education camps has a floor space between 4,500 and 8,000sqm. For example, Wusu City (Ili Prefecture) commissioned a Legal Systems Training School in July 2017 with a five-floor dormitory building featuring a floor space of 7,670sqm.[91] The Vocational Training Center commissioned by Shuimogou District in Urumqi City in April 2018 was to have a six-floor dormitory building with 7,300sqm for males, and a 4,700sqm building for females. These camps are essentially only medium-sized in terms of total floor space. However, assuming that 65 percent of the floor space is used for actual dorm space, they could easily intern between 2,500 and 3,900 persons when assuming a generous 2sqm dorm floor space per person, or two to four times as many under more crowded conditions (reports of detainees having to sleep on their sides, or taking turns sleeping, indicate a dorm floor space of 0.5 to 1sqm per person).

For the lower end minimum estimate of 812,000 in concurrent internment (900,000 currently or formerly interned), we would still estimate at least 100 large facilities that could hold 4,500 persons each, with the remaining smaller camps interning an average of only about 300 persons. At that capacity, these camps could literally consist of only one medium-sized (3,000 to 4,000sqm) building and still give those interned ample space. For example, the single-building (six-floor) internment camp photographed by the New York Times (Figure 28), located right next to the Hotan Legal System Education and Training Center (Figures 25-27) measures a total floor space of approximately 4,200sqm.

Figure 28: Single-building Hotan internment camp. Sources: Google Earth (37°14’56.37″N 79°50’6.04″E) and New York Times article (see footnote 87)

When considering all available sources of information, these new upper and lower limit estimates of 1.8 million and 900,000 respectively are eminently realistic, and appear to be an appropriate increase over previous estimates.

10. The Internment Campaign is Having a Dramatic Negative Social Impact
10.1 Declining Birth Rates

The dramatic adverse social impact of the internment campaign is reflected in a number of government statistics and reports. The first and perhaps most significant impact is reflected in drastically declining birth and net population growth rates. Until 2016, the combined net population growth rate of Kashgar and Hotan (weighted by population) consistently exceeded those of Xinjiang, and certainly the national average. In 2017, it fell by 31.9 percent compared to the previous year, falling substantially below the average of the previous 11 years (2005 to 2016). In 2018, the decline was far more dramatic, with net population growth falling by 72.3 percent compared to 2017, 81.1 percent compared to 2016, and 84.7 percent compared to the peak in 2013. Kashgar and Hotan’s net population growth was now substantially below that of the national average (2.58‰ versus 3.81‰).

Figure 29: Sources: Socio-economic development reports of each region shown.

A more detailed comparison of the birth rates within different regions in Xinjiang indicates that Xinjiang’s overall negative trend is mostly related to the dramatic declines in Hotan and Kashgar. Birth rates in regions where the shares of Turkic minority populations are much lower, such as Altay and Changji, declined only slightly. In Urumqi, birth rates even increased slightly. This confirms that drastic birth rate declines are largely restricted to Uyghur majority regions. These declines strongly correlate in both chronological and geographical terms with the internment campaign.

Figure 30: Sources: Socio-economic development reports of each region shown.

Whether death rates in Uyghur majority regions are an accurate reflection of the real situation on the ground is unclear. Death rates have not increased dramatically, but death rates as a percentage of birth rates more than doubled compared to the trend of the past 13 years. This is simply due to the fact that death rates increased slightly, while birth rates fell drastically. As a result, net population growth rates in these two Uyghur prefectures have plummeted.

Figure 31: Sources: Socio-economic development reports of each region shown.

10.2 The Socio-Economic Impact of the Internment Drive

Xinjiang’s unprecedented internment drive has literally torn families apart, depriving them of both intimate relationships and the labor that they relied on to make a living.

Below are several examples of households from Yarkand (Shache) County that are taken from local government lists of households that have fallen into poverty. Table 3 shows a young family with five children aged 3 to 14 years. The father was sentenced to prison, and the mother is in re-education (huzhu panxing, qizi shou jiao 户主判刑,妻子收教). As a result, their five children have become effectively orphaned.

Household UID Individual UID Address Name Gender Age ID number Description
58 217 古勒巴格镇巴格提坎村1组35号 阿布都热合曼·吐热普 male 36 653125198101****** Sentenced
218 尼亚孜布·吐尔孙 female 33 653125198312****** Re-education
219 阿卜杜巴斯提·阿布都热合曼 male 14 653125200306****** Minor
220 木巴热克·阿布都热合曼 female 10 653125200704****** Minor
221 阿布都拉·阿吧都热合曼 male 8 653125200910****** Minor
222 麦斯图热姆·阿布都热合曼 female 6 653125201110****** Minor
223 木叶萨尔·阿布都热合曼 female 3 653125201402****** Minor

Table 3: Government-issued spreadsheet listing households that are near or below the poverty threshold. Source: Source reference code 31702.

In the second example (Table 4), the head of household and only male (aged 33) is detained in re-education. His 29-year-old wife and four young children are left to fend for themselves. The husband’s detention into re-education is given as the official reason why this household is classified as poor.

The third example is similar, but in this three-generation household, 3 of 8 family members are interned. Among the interned are the relatively young grandparents (aged 48) and a young man, presumably the husband of one of the two young women, and father to one of the little girls. Only one male adult is left in the house. Since the young women appear to have young children to care for, that 23-year-old man is now the only obvious breadwinner for the 5 persons left in the house.

Individual UID Address Name Gender ID number Age Description
466 喀尔苏乡3村2组93号 艾麦提尼亚孜·伊则比拉 male 653125198402****** 33 Re-education
469 阿克柯孜·亚森 female 653125198812****** 29 Wife
467 阿依夏·艾麦提尼亚孜 female 653125201303****** 4 Minor
468 祖麦尔娅·艾麦提尼亚孜 female 653125201501****** 2 Minor
470 艾丽菲热·艾麦提尼亚孜 female 653125200905****** 8 Minor
471 祖丽非热·艾麦提尼亚孜 female 653125201104****** 6 Minor
541 喀尔苏乡3村7组395号 努尔敦·阿卜杜赛麦提 male 653125196907****** 48 Re-education
543 阿尔祖古丽·萨迪克 female 653125196906****** 48 Re-education
548 麦麦提·努尔敦 male 653125198905****** 28 Sentenced
545 布祖拉·沃拉木 female 653125199103****** 26 Adult
544 努尔比·扎克尔 female 653125199403****** 23 Adult
547 努尔艾合麦提·努尔敦 male 653125199407****** 23 Adult
546 阿依谢·麦麦提 female 653125201505****** 2 Minor
542 麦热孜亚·麦麦提 female 653125201309****** 4 Minor

Table 4: Government-issued spreadsheet listing households that are near or below the poverty threshold. Source: Source reference code 31702.

Table 5 shows the fourth example. Here, the father is in re-education, leaving four children (three of whom are still minors) in the care of his wife. Again, the father’s re-education internment is listed as the reason for the household’s poverty.

Individual UID Address Name Gender Age ID number Description
111 2村3组007 喀伍力·麦麦提 male 47 653125197105****** Re-education
114 帕太姆·奥斯曼 female 44 653125197405****** Wife
110 迪丽努尔·喀伍力 female 19 653125199903****** Adult
115 图尔荪·喀吾力 male 15 653125200305****** Minor
112 凯米尔丁·喀伍力 male 6 653125201201****** Minor
113 乃吉米丁·喀伍力 male 9 653125200910****** Minor

Table 5: Government-issued spreadsheet listing households that are near or below the poverty threshold. Source reference code: 34599.

Finally, the case shown in Table 6 exemplifies what has sadly become a fairly typical situation in Xinjiang. The middle generation (husband and wife) are interned or imprisoned, leaving the elderly grandparents to care for two very young grandchildren. The situation is further complicated by the fact that the grandmother is seriously ill. The family is described as “lacking labor force and finances”. Even when the young mother is released, the family will be dependent on state welfare for a long time, because the husband is scheduled to be in prison until the year 2030.

Household UID Individual UID Address Names Gender Age ID number Description
65 252 古勒巴格镇巴格提坎村2组41 阿布都赛麦提·库尔班 male 73 653125194406****** Grandfather
(household head)
253 麦热木•亚库普 female 63 653125195404****** Grandmother (seriously ill)
254 努尔艾力• 塞麦提 male 27 653125199004****** 13 year prison sentence
255 布热比•吐尔孙 female 22 653125199503****** Re-education
256 赛丽麦·努艾力 female 3 653125201409****** Minor
257 萨比热•努艾力 female 2 653125201511****** Minor

Table 6: Government-issued spreadsheet listing households that are near or below the poverty threshold. Source reference code: 31315.

These examples complement the overall internment-related statistics presented in the previous section. While not being statistically representative, they exemplify the real-world impact of the internment drive. These cases also confirm that the internment drive targets males, especially household heads. As a result, large numbers of families have been driven into poverty, leaving them dependent on the government for financial support.

The devastating economic impact of Xinjiang’s internment drive is also reflected in lists of government loan defaulters. Often, the reason given for default is the detention of a key breadwinner. For example, a list of loan defaulters from Pilal Township (皮拉勒乡) in Akto (Aketao) County, dated September 2018, shows that in 5 of 14 instances, the inability to repay is linked to internment. In 4 of these 5 cases, most of the funds are still in the bank. With loans issued in January and August 2017, the debtors were apparently detained before they could invest all the money.[92]

A particularly pertinent case comes from Yangaghlik Village (央阿格勒克村) in Azatbagh Township (阿扎提巴格乡), Yarkand (Shache) County.[93] There, a Uyghur farmer and head of a family was interned in 2017, just after the family was officially lifted out of poverty at the end of 2016. In October 2016, the farmer received a 40,000 RMB loan to purchase a piece of large agricultural machinery, as well as 2.65kg of Jujube seeds. After his detention, the expensive new equipment was no longer in use, and the loan could not be repaid as scheduled. In order to (again) rescue the household from poverty, the government had the family rent out their farming equipment and told the oldest son to get work. Subsequently, the local government marked this family as “poverty-alleviated by benefiting from policies” (yi tuopin; xiangshou zhengce 已脱贫;享受政策). Remarkably, the state did not let him go home in order to continue using his new equipment for farming, and so to repay the loan. In June 2018, the farmer applied to the government for assistance with the repayment of the loan and the related interest. In January 2019, he started to work in the Yarkand (Shache) County textile industrial park, earning only 800 RMB per month. Even worse, a form dated February 2019 shows that his son became disabled and unable to work in 2018 (at age 20), without stating a reason. However, it is clear that this family is now much worse off compared to three years ago.

Yet another depressing example comes from Awat Town (阿瓦提镇巴扎) in Azatbagh Township (阿扎提巴格乡), Yarkand (Shache) County (document dated April 2018). In a family of 5, the father and the oldest son were sent to re-education. Since one of the daughters married elsewhere, the wife was now left with her younger son to farm their 1.8 acres (11 mu) of land. No one repaired their home’s damaged water pipes. The document states that the younger son did not meet his obligations of helping the mother, and instead “begged for money”. The local government’s proposed “solution” to this situation was to conduct ideological education by “vigorously promoting the traditional virtues of the Chinese nation to respect the old and love the young, and enhancing the legal awareness and moral values of poor households.”[94]

These cases again shed light on the devastating real-world impact of the internment drive. They also show how local governments scramble to deal with the socio-economic fallout of this mass atrocity.

11. Conclusions

The classified telegram states that “the most important aspect” of the VTIC work is to “firmly establish a long-term warfare mindset”. Similarly, in an internal speech held in 2014 and leaked to the New York Times, China’s president Xi Jinping demanded that the thinking of Xinjiang’s minorities must be rewired through tougher measures, and that the viruses of their minds must be dealt with through “a period of painful, interventionary treatment.” He exhorted Xinjiang’s leaders to “wield the weapons of … dictatorship” “without hesitation or wavering”, and to show “absolutely no mercy.”[95]

Mr. Xi’s personal culpability for this unprecedented human rights atrocity in Xinjiang is now more evident than before, and this deserves particular attention. If there was ever any doubt that he explicitly condoned or perhaps even commanded the devastating policies in Xinjiang, that doubt is now removed. Mr. Xi bears the full responsibility for the unlawful internment of innocent citizens; for family separations, trauma, dehumanizing conditions and the inordinate mortality rates that result from his policies. He is also responsible for the torture, rape and murder that are either being perpetrated in an intentional and systematic fashion, or that would invariably result from the oppressive conditions created by the mass internment.

Along with other despots such as Hitler, Stalin, Mao Zedong or Kim Jong Un, Mr. Xi represents yet another historic example of the severe consequences for humanity when leaders wield ultimate and unchecked power. The ongoing human rights atrocity in Xinjiang is nothing less than a litmus test for the world’s most basic shared values. If the international community does not decisively respond to this, then it better stop claiming that it actually cares about human rights for all people. As a religious devotee worded it thousands of year ago: “Faith without works is dead.”[96]

Appendix: Chinese Terms for Vocational Training Internment Camps (VTICs)

In Chinese, “Vocational Skills Education Training Centers” (zhiye jineng jiaoyu peixun zhongxin 职业技能教育培训中心) are often abbreviated as “Education Training Centers” (jiao pei zhongxin 教培中心).[97] These are the only types of extrajudicial forms of internment whose existence Beijing has officially acknowledged, although it argues that they have a legal basis.

In Xinjiang, the abbreviated form “Education Training Centers” (jiao pei zhongxin 教培中心) very consistently refers to the “Vocational Skills Education Training Centers” (zhiye jineng jiaoyu peixun zhongxin 职业技能教育培训中心). An alternative term is “Education Training Centers” with all characters spelled out (jiaoyu peixun zhongxin 教育培训中心). Another common use is the term “Vocational Education Training Center” (zhiye jiaoyu peixun zhongxin职业教育培训中心 ) which leaves out the word “skills”.[98]

VTICs in Xinjiang can typically be distinguished from other vocational facilities by the ending “center” (zhongxin 中心).  If a facility or institution ends with the word “school”, as in “Vocational Skills School” (zhiye jishu xuexiao 职业技术学校), then it is typically not a VTIC. This distinction is for example evident from a document that lists the “Lop County Vocational Skills School” next to the “Education Training Center” (loupu xian zhiye jishu xuexiao he jiao pei zhongxin 洛浦县职业技术学校和教培中心).[99] However, the term “Education Training Center” (jiaoyu peixun zhongxin 教育培训中心) by itself is commonly used in Xinjiang and elsewhere for other types of training, including teacher training, cadre training, or for private learning institutions. It can denote a VTIC, but this must be established from the context.

Adrian Zenz is a Senior Fellow in China Studies at the Victims of Communism Memorial Foundation, Washington, D.C. (non-resident), and supervises PhD students at the European School of Culture and Theology, Korntal, Germany. He specializes in China’s ethnic minority policy, minority education systems, public recruitment (especially teacher and police/security-related recruitment), public bid documentation, domestic security budgets and securitization practices in China’s Tibetan regions and Xinjiang. He has authored Tibetanness under Threat (Global Oriental, 2013) and co-edited Mapping Amdo: Dynamics of Change (Prague: Oriental Institute, 2017). This paper is the substantially expanded peer reviewed version of an earlier working paper published in the Journal of Political Risk.

[1] https://www.bloomberg.com/news/articles/2019-04-19/how-china-is-defending-its-detention-of-muslims-to-the-world, https://www.bbc.com/news/blogs-china-blog-48700786.

[2] http://www.npc.gov.cn/wxzl/wxzl/2000-12/10/content_4509.htm, article 9.

[3] See for example http://www.gov.cn/gzdt/2009-10/19/content_1443436.htm or http://www.guoluo.gov.cn/html/892/185177.html.

[4] See for example http://www.gov.cn/gzdt/2009-10/19/content_1443436.htm and http://www.guoluo.gov.cn/html/892/185177.html.

[5] See https://www.ws.gov.cn/xwdt/bddt/20190119/i293954.html or http://archive.is/heMys, and http://www.kashi.gov.cn/Government/PublicInfoShow.aspx?ID=2976 or http://archive.is/pXZbj.

[6] See http://archive.is/rtpng or http://www.xjcbcr.gov.cn/info/1024/62397.htm and subsequent pages.

[7] Chinese: 职业技能教育培训中心把宗教极端思想从那些受到“三股势力”蛊惑的人的头脑中清除出去. Sources: http://www.xjfy.gov.cn/zwxx/001005/20181122/30bbf141-eae2-47e2-b233-b60ce14bbd13.html or http://archive.is/OowEo, or http://www.xj-agri.gov.cn/fslj/43101.jhtml or http://archive.is/THLVC, or http://www.xjkpx.gov.cn/xwzx/xjyw/20181123/i141354.html or http://archive.is/DBEwU, or http://www.xjks.gov.cn/Item/43288.aspx or http://archive.is/XyHq2, or http://www.xuegong.cug.edu.cn/info/1051/6341.htm or http://archive.is/XAMeA

[8] https://web.archive.org/web/20190630195027/http://www.xinyuan.gov.cn/info/egovinfo/1001/common/inf_content/xy022-02_A/2017-0526003.htm or http://archive.is/ARyEg. Compare this with the similar statement “伊犁州关于开展“清脑净心、扶正祛邪”专项行动的实施方案” from http://www.qxjdr.com/shuma/5447.html or http://archive.is/iNOU1.

[9] For a detailed discussion refer to the author’s full academic paper on Xinjiang’s re-education campaign at https://www.tandfonline.com/doi/full/10.1080/02634937.2018.1507997 or https://osf.io/preprints/socarxiv/4j6rq/

[10] https://www.tandfonline.com/doi/full/10.1080/02634937.2018.1507997 or https://osf.io/preprints/socarxiv/4j6rq/

[11] http://xjtzb.gov.cn/2016-03/14/content_509505_3.htm or http://archive.is/YEJdd.



[14] https://jamestown.org/program/the-learning-curve-how-communist-party-officials-are-applying-lessons-from-prior-transformation-campaigns-to-repression-in-xinjiang/

[15] https://www.tandfonline.com/doi/full/10.1080/02634937.2018.1507997 or https://osf.io/preprints/socarxiv/4j6rq/

[16] http://www.wlmqx.gov.cn/fxj/xs/15706.jhtml or http://archive.is/hD9qO

[17] http://www.ouqiao.net/13/2018-10-18/8155.html or http://archive.is/jBOLp

[18] http://www.xjkjt.gov.cn/2017/09/15/tpzshi/12167.html or http://archive.is/9xRGw

[19] http://www.xjb.cas.cn/ddjs2016/201710/t20171008_4868178.html or http://archive.is/2Ssdt.

[20] http://www.xjvu.edu.cn/news/150502026779635366.html or http://archive.is/4lYyi, and http://xjlyt.gov.cn:88/tpxw/1109.htm or http://archive.is/kLA78.

[21] For example: http://m.chinaxinjiang.cn/view-783e859cacc34402bbf03494891b5041.html or http://archive.is/cDGd0; http://www.xj-agri.gov.cn/nongyeyw/36010.jhtml or http://archive.is/jhcbS; http://www.xj-agri.gov.cn/nongyeyw/35329.jhtml or http://archive.is/t81aN

[22] https://web.archive.org/web/20190807205819/http://www.fk.gov.cn/wcm.files/upload/CMSfk/201806/201806060532047.doc. See also http://finance.sina.com.cn/roll/2017-10-10/doc-ifymrcmm9963055.shtml or http://archive.is/9Cpws; also http://www.cnbxgs.net/?/bwc/aqcs/25818.htm or http://archive.is/9Cpws.

[23] http://www.xjfzb.com/contents/240/117002.html, http://jyj.beijing.gov.cn/ywgknew/2377.htm, http://news.cjn.cn/sywh/201804/t3198456.htm

[24] https://chinese.uhrp.org/article/358558867

[25] http://www.globaltimes.cn/content/1141882.shtml or http://archive.is/7mLZX

[26] See https://www.tandfonline.com/doi/full/10.1080/02634937.2018.1507997 or https://osf.io/preprints/socarxiv/4j6rq/ for details.

[27] http://www.qianlima.com/zb/detail/20170707_60399745.html (link dead) and https://web.archive.org/web/20180617184616/http://www.bidchance.com/info.do?channel=calgg&id=21177553 and https://web.archive.org/web/20180617184616/http://www.bidchance.com/info.do?channel=calgg&id=21056414 and https://web.archive.org/web/20180906113745/http://www.dlzb.com/d-zb-1685922.html

[28] http://www.xj.cei.gov.cn/info/10856/368183.htm (link dead). See also https://www.tandfonline.com/doi/full/10.1080/02634937.2018.1507997 or https://osf.io/preprints/socarxiv/4j6rq/. Chinese state TV footage of a “vocational training center” model classroom showed numerous cameras and microphones on the walls and the ceiling (see http://tv.cctv.com/2018/10/16/VIDEVvr9aq34SsDMrB6IRGnh181016.shtml).

[29] http://ztb.xjjs.gov.cn/xjweb/ZtbInfo/ZBGG_Detail.aspx/?InfoID=811ce665-2956-4576-8340-5b4bcf2a8868&CategoryNum=004001001

[30] http://www.dlztb.cn/news/201804/25/304792.html or http://archive.is/Fb9NP

[31] https://www.okcis.cn/20171009-n2-20171009145031635272.html or http://archive.is/idmwu.

[32]https://web.archive.org/web/20181204024839/http://kashi.gov.cn/Government/PublicInfoShow.aspx?ID=2963. Compare http://www.kashi.gov.cn/UploadFiles/News/2019/3/201903281646572803.zip. “Demanded by Chen Quanguo”: http://wemedia.ifeng.com/63532661/wemedia.shtml or http://archive.is/Qq5V5

[33] 把“防闹事”摆在突出位置的同时,统筹抓好防脱逃、防疾病、防火、防地震等工作措施. The five measures are: disturbance prevention, escape prevention, disease prevention, fire prevention, earthquake prevention. Source: http://wemedia.ifeng.com/63532661/wemedia.shtml or http://archive.is/Qq5V5

[34] https://web.archive.org/web/20190313014950/http://www.xjcl.gov.cn/__local/7/DC/41/5B59113406996B0C8B5C3F0CD27_E8DAF9BC_B3A00.xls. Download page: http://www.xjcl.gov.cn/info/1108/52843.htm or http://archive.is/WqTtS

[35] http://wemedia.ifeng.com/63532661/wemedia.shtml or http://archive.is/Qq5V5

[36] http://www.moj.gov.cn/organization/content/2017-06/05/zzbsjxw_4479.html or http://archive.is/9CrDb

[37] http://lcqfy.chinacourt.gov.cn/public/detail.php?id=2583 or http://archive.is/8zMCu

[38] https://www.rfa.org/english/news/uyghur/detentions-11012018100304.html

and https://chinese.uhrp.org/article/358558867


[40] http://www.chinagwy.org/html/gdzk/xinjiang/201711/87_219677.html or http://archive.is/TDNPh.

[41] http://www.kashi.gov.cn/Item/47112.aspx or http://archive.is/hEYxE

[42] https://web.archive.org/web/20190313014950/http://www.xjcl.gov.cn/__local/7/DC/41/5B59113406996B0C8B5C3F0CD27_E8DAF9BC_B3A00.xls. Download page: http://www.xjcl.gov.cn/info/1108/52843.htm or http://archive.is/WqTtS.

[43] http://www.xjcl.gov.cn/info/1079/60125.htm or http://archive.is/V0TGo

[44] https://www.nytimes.com/2019/03/02/world/asia/china-muslim-detention-uighur-kazakh.html

[45] Source reference code: 27271.

[46] GPS coordinates: 43°59’52.5″N 87°31’00.8″E

[47] The author verified that bus number 3003 does indeed go right past the location of the compound, using the main road shown in figure 2.

[48] https://www.youtube.com/watch?v=6sUEek-u14w, the relevant video section is from 0:15 to 0:23. The video shows the same buildings, walls and watchtowers as on the Google Earth images. Note how the yellow-black striped wall that separates each side of the road is also visible on the satellite image.

[49] Source reference code: 28944.

[50] http://www.xjcl.gov.cn/__local/F/9E/CE/EC441FEF3583C7AC2B0E331E5BB_1541E1B9_19DCF.wps; http://www.jmser.gov.cn/wcm.files/upload/CMSjmser/201807/201807090130047.doc. This must be distinguished from the related aministrative bureau for vocational skills training (zhiye jineng peixun fuwu guanliju 职业技能培训服务管理局).

[51] http://www.xinjiang.gov.cn/2017/03/30/128831.html or http://archive.is/J8jYB. Compare http://www.tkx.gov.cn/FileDownload.do?affixid=11063

[52] http://www.xinhuanet.com/politics/2018-11/05/c_1123667476.htm or http://archive.is/ZwWU0

[53] http://kashi.gov.cn/Government/PublicInfoShow.aspx?ID=3059 or http://archive.is/ZK8Os

[54] The document ends with the signature line: 阿克苏地区司法局(教培局). Source: https://web.archive.org/web/20190621120233/http://www.modestdiva.com/module/download/downfile.jsp?classid=0&file name=1807241609317428844.pdf

[55] http://ws.xjkunlun.cn/rdxw/2018/2735547.htm or http://archive.is/rRKLC

[56] See e.g. https://web.archive.org/web/20190313014950/http://www.xjcl.gov.cn/__local/7/DC/41/5B59113406996B0C8B5C3F0CD27_E8DAF9BC_B3A00.xls

[57] http://www.kashi.gov.cn/Government/PublicInfoShow.aspx?ID=2976 or http://archive.is/pXZbj. The original phrase is concealed in the document as “教培局:负责各◇◇◇◇学员参保缴费工作”. Compare also this document, where this fact is confirmed, and the phrase “教培中心” is not concealed but spelled out: http://www.xjks.gov.cn/Item/43426.aspx or http://archive.is/cmjoQ.

[58] http://www.akss.gov.cn/xjcms/front/s448/xxgk/20181123/i105526.html or http://archive.is/J2cDW

[59] The distinction between “vocational skills training” that mainly serves “transformation through education” purposes and genuine (professional) “vocational educational” (zhiye jiaoyu 职业教育) based on vocational skills institutions or schools (zhiye jishu xuexiao 职业技术学校) is confirmed by an analysis of Xinjiang’s budgets and spending patterns. In 2017, the region’s spending on “vocational education” (zhiye jiaoyu职业教育), which is administered by the education system, was actually below the national average on a per capita basis. In contrast, spending on the justice system (sifa xitong 司法系统) amounted to over three times the national average (7.9 versus 2.5 per capita). Between 2016 and 2017, Xinjiang’s spending on “vocational education” actually decreased by 7.1 percent (in a sample of ethnic minority regions by 3.3 percent), while spending on the justice system rose by 118 percent (in the ethnic minority region sample by 235 percent). The explanation for this is simple: just like the PRC’s former re-education through labor system, Xinjiang’s re-education campaign appears to be managed by the Ministry of Justice, administered by the public security agencies, and funded largely out of the budgets of these same authorities. Construction and procurement bids confirm these two agencies as the two main issuing authorities. Related expenditures are accounted under domestic security spending, not education system spending. See https://jamestown.org/program/xinjiangs-re-education-and-securitization-campaign-evidence-from-domestic-security-budgets/

[60] https://www.theglobeandmail.com/world/article-i-felt-like-a-slave-inside-chinas-complex-system-of-incarceration/

[61] See https://www.tandfonline.com/doi/full/10.1080/02634937.2018.1507997 or https://osf.io/preprints/socarxiv/4j6rq/.

[62] Pils, E. 2017. “Human rights in China”.

[63] For example, Mirigul Tursun (although she was in different facilities):   https://www.rfa.org/english/news/uyghur/detentions-11012018100304.html; Gulbahar Jelil: https://supchina.com/2018/12/05/starving-and-subdued-in-xinjiang-detention-centers/

[64] http://www.xjdaily.com.cn/tsnb/1150049.shtml or https://web.archive.org/web/20180827152731/http://news.163.com/14/1118/10/ABB06EE800014AED.html

[65] Source reference code: 18204

[66] http://www.xjcbcr.gov.cn/interact/mail_content.jsp?urltype=leadermail.LeaderMailContentUrl&wbtreeid=1322&leadermailid=1389 or http://archive.is/yDyEu


[68] Based on the 2015 mini census (2015年1%人口抽样调查) for Bachu County and Xinjiang. The population aged 0-17 years was estimated based on the 0-14 year cohort from the 2015 mini census and the 15-19 year cohort (multiplied by 0.5) from the 2010 census. Rapid population growth in Uyghur majority regions means that using 2015 census data makes for a much more accurate estimate than solely relying on the 2010 census data.

[69] http://www.zgtks.gov.cn/shishixinwen/zwdt/2016-01-05/41697.html or  http://archive.is/z9eNV

[70] Source reference code: 18305.

[71] Source reference code: 18761.

[72] Source reference code: 21963.

[73] https://web.archive.org/web/20190926030535/http://www.xjkjt.gov.cn/content/downloadAttachment.shtml?id=129&attachmentUrl=/upload/2018_5/814080

[74] Source reference code: 22735.

[75] http://archive.is/gOPNU (index: http://www.xjcbcr.gov.cn/open/zdxxgk/shbz/2.htm)

[76] Subsidy term: 职业教育培训中心学员伙食费补助https://web.archive.org/web/20190706203806/http://www.xjcz.gov.cn/yjsxxgk/zfyjs/zfys/zfys2019/201901/P020190403725265048327.xlsx

[77] See e.g. https://supchina.com/2018/12/05/starving-and-subdued-in-xinjiang-detention-centers/

[78] http://www.chinadaily.com.cn/hqzg/2007-07/18/content_5438448.htm or http://archive.is/95wfz. Compare this figure with the food budgets or allowances of private companies, for example 3.50 RMB per warm meal: http://www.stcbao.com/article/canbiao12.html.

[79] Sources: http://www.kashi.gov.cn/Item/47217.aspx or http://archive.is/SGRUr, and the 2010 and 2015 census age cohort data (see footnote 64 for the calculation method).

[80] https://www.rfa.org/english/news/uyghur/detentions-01222018171657.html

[81]  Various conversations and message exchanges conducted by the author in the second half of 2018.

[82] See e.g. https://www.aspi.org.au/report/mapping-xinjiangs-re-education-camps, https://www.reuters.com/investigates/special-report/muslims-camps-china/, https://www.bbc.co.uk/news/resources/idt-sh/China_hidden_camps

[83] https://jamestown.org/program/xinjiangs-re-education-and-securitization-campaign-evidence-from-domestic-security-budgets/

[84] https://www.tandfonline.com/doi/full/10.1080/02634937.2018.1507997 or https://osf.io/preprints/socarxiv/4j6rq/.

[85] See https://osf.io/preprints/socarxiv/4j6rq/, p.28-29

[86] https://jamestown.org/program/xinjiangs-re-education-and-securitization-campaign-evidence-from-domestic-security-budgets/

[87] See See https://medium.com/@shawnwzhang/satellite-imagery-of-xinjiang-re-education-camp-5-%E6%96%B0%E7%96%86%E5%86%8D%E6%95%99%E8%82%B2%E9%9B%86%E4%B8%AD%E8%90%A5%E5%8D%AB%E6%98%9F%E5%9B%BE-5-1620c20a8deb and https://www.nytimes.com/2018/09/08/world/asia/china-uighur-muslim-detention-camp.html

[88] See e.g. the Shahit database, which contains Hui victims despite the fact that the Hui typically do not have extensive links to communities abroad, and that therefore their internment is less likely to be reported.

[89] Because the census only gives ages per county by 5-year cohorts, the adult population was calculated by adding 40 percent (2/5) of the age cohort 15-19 years to the population aged 20 years and higher.

[90] 2018 XUAR population figure source: XUAR 2018 socio-economic development report

[90b] See https://www.jpolrisk.com/karakax/.

[91] See https://osf.io/preprints/socarxiv/4j6rq/, Table 1 (list of government bids), also for subsequent references to bids.

[92] Source reference code: 25611

[93] Source reference code: 25874.

[94] Source reference code: 25947.

[95] https://www.nytimes.com/interactive/2019/11/16/world/asia/china-xinjiang-documents.html

[96] https://www.biblegateway.com/passage/?search=James+2%3A14-26&version=ESV

[97] http://www.alt.gov.cn/zwxx/001002/20190321/7f49da54-0cb5-47ad-b486-d07bb4060a8a.html or http://archive.is/QPael. Compare http://www.scio.gov.cn/ztk/dtzt/39912/40016/40030/Document/1649854/1649854.htm or http://archive.is/b5j3L

[98] http://www.akss.gov.cn/xjcms/front/s448/xxgk/20181123/i105526.html or http://archive.is/J2cDW. Yet another term is职业技能教培中心, e.g. in https://web.archive.org/web/20180813115300/http://www.kashi.gov.cn/Government/PublicInfoShow.aspx?ID=2851

[99] http://www.xjlpx.gov.cn/nzcms_nzweb/nzcms_up/attached/file/20190411/20190411130965606560.pdf



On February 7, 2022, the section referring to the release data from Shahit was removed, because Shahit release data does not only pertain to releases from internment facilities. Instead, this estimate was based on the leaked internal document “Karakax List”, which states release information for Karakax County for 2018 and approximately the first half of 2019. This does not change the total internment estimate range.