Paper short abstract:

Drawing on ethnographic research on smart city initiatives in China, I demonstrate how regional governments and startup companies use the logics of machine learning as a heuristic to understand how the state crowdsources policy solutions, as well as their own role within such a system.

Paper long abstract:

It is a longstanding but controversial notion that Chinese society works through top-down command chains that impose totalizing homogeneity. Anthropologists have challenged an univocal picture that overstresses hegemonic power of by exploring themes of resistance (e.g. Weller 1994) and an informal “second society” that operates beyond the state (e.g. Yang 1989). In this paper, I add nuance to this debate by providing local perspectives on how the Chinese state operates. Drawing on fieldwork I conducted in China on the local implementation of national smart city mandates, I show how regional governments and startup companies understand the operation of the Chinese state as akin to a machine learning algorithm that solves problems by parsing large data sets without explicit programming. Seeing themselves as policy-generating nodes within a nationwide machine learning assemblage, these actors understand this computational analogy to be simultaneously a reinterpretation and a seamless continuation of the late paramount leader Deng Xiaoping’s philosophy: “It doesn’t matter whether a cat is black or white, as long as it catches mice, it is a good cat.” My analysis clarifies the role of the market economy in Chinese governance under Capitalism with Chinese Characteristics. It also dispels a popular misconception about so-called Chinese “collectivism”: instead of a bland “copy-and-paste” homogeneity, Chinese governance relies on the constant collective production of an abundance of diversity, in effect enlisting local governments to form what AI practitioners call generative adversarial networks (GANs).