We discuss the tensions between intensionality and extensionality of spatial observations in distributed systems, showing that there are natural models where extensional observational equivalences may be characterized by spatial logics, including the composition and void operators. Our results support the claim that spatial observations do not need to be always considered intensional, even if expressive enough to talk about the structure of systems. For simplicity, our technical development is based on a minimalist process calculus, that already captures the main features of distributed systems, namely local synchronous communication, local computation, asynchronous remote communication, and partial failures.