However, existing unsupervised skill discovery methods often learn entangled skills where one skill variable simultaneously influences many entities in the environment, making downstream skill chaining extremely challenging.

Despite growing interest in understanding and predicting the long-term behavior of such systems, recent studies have revealed a wide array of negative results, demonstrating the elusiveness of game dynamics.

However, the evaluation process presents substantial challenges.

We revisit this scenario as an experimental case study in Section 6.