Decentralised policymaking in China is often cited as a key success factor in economic reform and authoritarian resilience. Although the existing literature presents policy diffusion as a technocratic process where socially optimal policies diffuse, many examples exist where the reverse is true or where the central government sanctioned local innovation but the policy diffuses to other places regardless. The author contends that policy diffusion in authoritarian regimes should be understood as a political process, where local officials serve as policy entrepreneurs, rather than a technocratic one. Subnational officials do not respond uniformly to either incentives from the central government or local pressure, but rather adopt experimental policies as a strategy learned from other successful officials. Policy experimentation has emerged as a strategy for officials desiring either career advancement or security, resulting in an S-shaped curve of policy diffusion characteristic of a learning process whereby a few initially innovate but others quickly adopt the experiment once viewed as successful.