State representation is the encoding scheme used to capture the relevant information about a system or environment at a given point, transforming raw observations into a form suitable for prediction, planning, or synchronisation. Choices range from hand-crafted low-dimensional feature vectors to learned latent embeddings produced by an encoder network, and the choice materially affects downstream sample efficiency and generalisation. It is a prerequisite for constructing a world model and for synchronising state across distributed or networked simulations.