top of page


Untitled design (9).png

A Uyghur woman holds her ID up while a police officer takes her photo at a checkpoint in Ürümchi, Northwest China in 2018. Image courtesy of Yael Grauer.

Black Logo Mark.png

May 2021

There were three lanes at the checkpoint in Turpan—an old Uyghur-majority oasis city on the northern reaches of the Taklamakan Desert in Northwest China. Two lanes were fitted with turnstiles, ID and face-scanners, metal detectors, and “data doors” which matched the SIM card of cell phones to the IDs of individuals. The third lane, on the far left, went through a simple metal gate. It was opened by a Uyghur police assistant, who looked at the faces of the people lined up behind it. Only people who appeared to be Han were permitted to exit through the third lane. They did not show their IDs to the police assistant. The ethno-racial phenotypes of their faces were enough. Speaking in Uyghur to people around me I noted that two of the lanes were for “native” people (in Uyghur, yerlik) while the other lane was for settlers. Continuing, I said, “I’m not native or Han, so which line should I go through?” The Uyghur women next to me smiled slightly—a mix of embarrassment, nervousness and irony flashing across her face—at what my question implied. The framing of my question, something that is not often said out loud in public, made clear that the natives were being subjected to scans because of their ethno-racial difference, their belonging to this colonized Islamic landscape in Northwest China. 


She suggested I go through the “native” line. When I got to the front of the line a few minutes later, I told the Uyghur police assistant in Uyghur that I did not have a Chinese ID, I just had a passport. Responding to me in Uyghur, he said he would have to register me manually. As we walked to the police station he asked where I was from and how I learned Uyghur. I told him I was an American anthropologist who had lived in the Uyghur region for two years and that I had learned to speak Uyghur in the capital city, Urumchi. He said he had studied English when he was in school, but he had forgotten most of it because he had no one to practice with. As we approached the door he asked abruptly if I also spoke Mandarin Chinese. I responded in Chinese, “Of course.” As we entered the police station, I understood why he asked this question. We were entering a Chinese-speaking world. He led me to a Han officer and explained that I needed to be manually registered. The woman behind the desk scanned my passport picture and raised her smartphone to scan my face. I asked why she needed to scan my face. She responded, “It’s just to keep you safe.”

"In regions of contemporary states where such totalitarian infrastructures are implemented, they produce dispositions or patterns of propensity that arbitrate the possibilities of life itself."


As this was happening, a pale-faced Uyghur young man was led into the station behind me. His ID had caused an alarm as he went through the checkpoint. His hands were shaking and he stuttered while he tried to explain why his ID had set off the system. I did not have time to stay and find out what was to be done with him. My privilege as a passport-carrying foreigner meant that a pathway through the flexible enclosures of the policing system was opened up for me, just as they were being opened for Han settlers at the checkpoint. At the same time, the face and ID of the young man had set off the alarms of the system and the walls were closing in around him. 


Michel Foucault (1997) has argued that the enframing effect of architecture is what makes it effective in the enactment of power relations at the grass-roots or capillary level of society. Building on this approach, the anthropologist Allen Feldman (1991) argues that power is not in fact “distributed” from a center of power held in reserve as much it is enacted through a “metonym of doing.” Power, defined as the ability to affect another and be affected, is enacted not by a unitary state, but rather discrete acts of doing within the architecture of a state.  It is here at the jurisdictional boundaries of state institutions—for instance at a checkpoint in Turpan—that the material experience of the state is given structure and significance; movement is regulated; the environment is controlled. The structures of walls and gates, and the infrastructures of surveillance, open up certain forms of movement while foreclosing others. The policing and carceral systems that surveillance systems both symbolize and support, interpellate people as subjects, reinforcing the protections of settler society while undermining the autonomy of the colonized. In regions of contemporary states where such totalitarian infrastructures are implemented, they produce dispositions or patterns of propensity that arbitrate the possibilities of life itself.

"The police documents list violations such as being part of a Quran study group on the social media app WeChat or contacting family members who lived in Muslim majority countries as reasons why the person was detained. In many cases when these activities took place they had not yet been criminalized."


Since late 2019 I have been combing through a 52 gigabyte internal policing database from the capital of the Uyghur region, Urumchi. These police files obtained by The Intercept contain thousands upon thousands of reports of Uyghurs and others who were stopped at checkpoints just like the one I went through in Turpan in 2018. I have looked at thousands of images of young Uyghurs staring into the camera of a police officer as they are investigated. Like the young man who was led into the station behind me at the Turpan checkpoint they are moments away from finding out if digital scans of their devices and social network will determine if they are “normal” or “untrustworthy.” The police documents list thousands of names, ID numbers, and geolocations of Uyghurs who have been taken away either to prison or a sprawling reeducation camp system. In the neighborhoods, where I have the best police data between 5 and 10 percent of ethnic minority adults were detained (Byler 2021). Often their crimes, or what might be better labeled “pre-crimes” -- since they are described in state documents as “not reaching the level of criminality” -- had to do with algorithmic assessments of digital activity. The police documents list violations such as being part of a Quran study group on the social media app WeChat or contacting family members who lived in Muslim majority countries as reasons why the person was detained. In many cases when these activities took place they had not yet been criminalized. The total infrastructures of contemporary control collapse the past into the present, even as they mediate the individual choices of the future. As the anthropologist Sareeta Amrute has noted racialized algorithms leave a long tail, affecting social life across multiple domains (2020). In the case of totalizing infrastructures, the long tail transcends temporal dimensions. Past behavior is now being used to diagnose current status: trustworthy, normal, untrustworthy.  


 A Uyghur man looks into a police officer’s camera at a checkpoint in Ürümchi, Northwest China in 2018. Image courtesy of Yael Grauer.

According to the Chinese policing literature, the system as it is operationalized across the Uyghur region utilizes a combination of counter-insurgency theory or COIN and a type of so-called predictive policing or Countering Violent Extremism (CVE) (Byler 2019). Technology enabled “full spectrum intelligence” is used to assess the population. This system builds on a grassroots policing system which made civil ministry workers responsible for the behavior of people in units of 100-1000 households. These neighborhood watch units, often referred to as shequ emerged out of the Maoist period, but have been elaborated further through a grid management system that allowed Chinese authorities to consolidate their power over the past several decades. With the arrival of digital forensics and biometric assessment tools in the mid-2010s state workers in the ideological “war zone” of Northwest China began to scan phones to determine who was susceptible to untrustworthy thoughts and to identify people on watch lists due to their untrustworthy relatives. In order to implement this system of control, the state has hired as many as 90,000 police assistants to monitor nested checkpoints and conduct spot checks of phones and IDs (Byler 2021). The system moves from public space, to domestic space, to the digital behaviors of individuals. 

"Ultimately then the state authorities and technologists who write the manuals and the code determine the calculus of this system."


The case of technology-guided totalitarian statecraft in Northwest China demonstrates that performances of power are concentrated in discrete spaces and times such as checkpoints where police assistants are trained to sort the population into Muslim and non-Muslim. Or as part of a system of periodic home visits where state workers inspect Muslim homes using a scripted checklist.  Or in scans of social media accounts using the algorithms of a digital forensics tool, to fulfill the intelligence quotas they are given by leaders higher in command. Ultimately then the state authorities and technologists who write the manuals and the code determine the calculus of this system. Rather than sources of power becoming simply more decentralized, in this context settler colonial relations of domination, and their performance, are more concentrated, just in self-replicating, modular forms in the space of checkpoints, the scripts of home inspections, and the interfaces of phone scanners. Studies by scholars such as Brian Jefferson (2020) have shown that these types of technologies are not limited to the Chinese case. Many of them are also used by policing agencies and private companies in places like the United States and Europe to reinforce border logics throughout democratic states (Sanchez Boe 2021). 


What makes the infrastructures of control in Northwest China different from systems deployed elsewhere in the world is ultimately not the technologies themselves but rather the scale and political system that regulates the technologies. The sheer scale and density of the deployment of checkpoints and biometric data collection in Northwest China is unparalleled elsewhere in the world. This material difference is matched by a political difference that centers on state-controlled media, the size of the surveillant population, and the near total absence of legal protections for the surveilled—who are deemed outside civil protections due to their proximity to the figure of the “terrorist.” The Muslims of Northwest China are a population of close to 15 million people—nearly three times the size of the Palestinian population in the West Bank, Israel and Gaza—making them the largest watch-listed population in the world, unable to travel freely or enjoy many of the limited freedoms of the majority of people in China. The technologies used in border policing, the image and face-recognition algorithms of Clearview and Palantir, the behavior analytics of Google and Facebook, do similar work to the tools used in Northwest China. The primary difference is their intensification, and the political regime that accesses their harvested data.


To reprise Allen Feldman’s framing (1991) of autonomized political violence in Northern Ireland, infrastructural power embeds power in the performance of both human and technical agents. In Northwest China, advanced technology systems allow infrastructural power to be programmed by state authorities and technologists and instantiated through the performance of state workers and their technical assessment tools. Information infrastructure when weaponized in this way turns the built environment, state, and corporate power into a single structure of continuous sorting and interrogation. The technical totalitarian systems of the present function with an automated intimacy that moves from chips in smartphones to phenotypes of faces, to datasets and watchlists, and back again. In this sense, they transcend scale, and produce a concentrated practice of power that becomes the context for its own reproduction.  


Sareeta Amrute, “Bored techies being casually racist: race as algorithm” (Science, Technology, & Human Values 45, no. 5 2020: 903-933).

Darren Byler, “Preventative Policing as Community Detention in Northwest China” (Made in China Journal, 25 October, 2019,


Darren Byler, “Chinese Infrastructures of Population Management on the New Silk Road” (Wilson International Center for Scholars, 2021).

Allen Feldman, Formations of violence: The narrative of the body and political terror in Northern Ireland (University of Chicago Press, 1991).

Michel Foucault, Security, territory, population: lectures at the Collège de France, 1977-78 (New York: Springer, 2007).

Carolina Sanchez Boe, The Undeported: The Making of a Floating Population of Exiles in France and Europe (Rowman & Littlefield Publishers, 2021).

Brian Jefferson, Digitize and punish: Racial criminalization in the digital age (University of Minnesota Press, 2020).


Darren Byler is an incoming Assistant Professor of International Studies at Simon Fraser University in Vancouver, British Columbia and a post-doctoral research fellow in the China Made Project at the Center for Asian Studies at the University of Colorado, Boulder. He is the author of a forthcoming ethnography titled Terror Capitalism: Uyghur Dispossession and Masculinity in a Chinese City (Duke University Press 2021) and a narrative-driven book titled In the Camps: China's High-Tech Penal Colony (Columbia Global Reports 2021). His current research interests are focused on infrastructure development and global China in the context of Xinjiang and Malaysia. He received his Ph.D in anthropology from the University of Washington.

Cite this article as: Byler, Darren. July 2021. 'Totalizing Infrastructures: Statecraft in the Uyghur Penal Colony'. Today's Totalitarianism.


White Logo Black Background.jpg
bottom of page