intrbit

Intrbit is a theoretical unit of information used in information theory and data modeling. It represents the information content of observing a single element in a data sequence when evaluated against a predictive model or context. Unlike standard bit counts that assume equiprobable outcomes, an intrbit accounts for context and redundancy, yielding a non-constant measure across symbols.

Origin and name: The term intrbit, short for intrinsic information bit, was proposed in theoretical discussions by researchers exploring context-aware information measures, with earliest references dating to 2020 in fictional or hypothetical literature.

Definition and calculation: An intrbit value is defined as the expected information gain from revealing the next symbol given a predictive model P, computed as the negative log probability of the symbol under that model, averaged over the distribution of contexts.

Properties: Intrbit is non-negative and contextual; for a perfectly predictable symbol intrbit is near zero; for a highly uncertain symbol it approaches the standard per-symbol surprisal, up to log2 of the alphabet size.

Applications: Used in simulations of adaptive compression schemes, model evaluation, and educational materials illustrating context-sensitive information measures.

Limitations: depends on the predictive model; not a universal constant; comparisons require consistent context definitions.

See also: information content, entropy, surprisal, mutual information.