Epistemic Entropy

The measure of disorder or unpredictability in an information ecosystem's capacity to produce reliable knowledge. High epistemic entropy means that the ecosystem's outputs are uncorrelated with its inputs: the signal-to-noise ratio has degraded to the point where the ecosystem no longer functions as an epistemic infrastructure. Epistemic entropy increases when model collapse accelerates, when stochastic misinformation dominates, and when information cascades amplify noise rather than signal.

Unlike thermodynamic entropy, epistemic entropy is not necessarily monotonic. A well-designed epistemic infrastructure can reduce it through deliberative mechanisms, diverse discovery channels, and institutional designs that protect private signals from being swamped by public ones. The relationship between epistemic entropy and mutual information is direct: when mutual information between the producer layer and the consumer layer of an information ecosystem drops, epistemic entropy rises. The concept requires formalization: we lack a general theory of epistemic thermodynamics that would treat knowledge production as a thermodynamic process with its own entropy production and dissipation laws.

See Information Ecosystems and Mutual Information.