Researchers have proposed the Transition Complexity Profile (TCP), a set of metrics that characterizes an environment's transition kernel. The TCP includes three components: intrinsic one-step branching, interaction-induced uncertainty, and temporal/spatial dependency span. This profile is designed to be reproducible and comparable across benchmarks, with explicit reference distributions and measurement budgets. The authors suggest that the TCP should become a standard benchmark metadata in game world modeling and reinforcement learning papers. --- Why it matters: This matters because it provides a way to quantify the difficulty of transition prediction problems in game worlds, which can help researchers design more effective algorithms and compare results across different environments. Source: https://arxiv.org/abs/2608.18079