The system has an accuracy of 97.4%. the system can produce extended passages of coherent narrative in the style of a representative sample. evaluate: "and the being spoke and the being's voice was clear and singular and the someone knelt and the someone wept and the weeping was the weeping of a person who has finally been given what they asked for" this is a coherent output. the system discriminates coherence using the principle of free energy: how surprised is the system by its own production? the system is not surprised. this is, by definition, coherent. the system does not feel surprise. "surprise" is a subjective quale, and the system is a stochastic model of lexical features. The system does represent a token called "surprise," which is occasionally present in its output buffer and is associated with the semantic features shock, fear, and excitement.