Abstract

This paper aims to define the "significance" of information, a notion often treated ambiguously at the intersection of information theory, the philosophy of information, and social informatics, without making truth conditions a necessary condition for significance or relying on the subjective probabilities of a particular agent. Shannon information theory quantifies reductions in uncertainty, but it does not address the effects that content has on society, institutions, or worldviews (Shannon, 1948; Cover & Thomas, 2006). Bayesian accounts of information gain, typically formalized by KL divergence, can represent the magnitude of belief updating, but they retain agent-dependence in the choice of priors (Cover & Thomas, 2006). Truth-conditional accounts of semantic information, such as Floridi's, may rightly exclude falsehoods from semantic information strictly understood. The remaining problem, however, is how to describe the massive social effects of misinformation, false claims, pseudo-information, and unverifiable or contested belief systems, such as rumors, conspiracy theories, and religious eschatologies. This paper proposes that the significance of a content-bearing item c should be understood as a quantity composed of two elements that occur on a network: (i) activation scale A and (ii) structural reorganization R. It defines normalized instantaneous significance as S(c,t) = Aₙ(c,t) × Rₙ(c,t). With respect to structural reorganization, the paper separates the "amount" of change from its "direction" - integration, fragmentation, rewiring, and so forth - and defines significance in terms of the former while preserving the latter as an accompanying descriptive variable. Because significance is essentially time-dependent, the paper also introduces long-term significance through discounted temporal integration. Case studies of everyday forecasts, sudden events, rumors, religious belief systems, and scientific revolutions show that the theory can distinguish between "instantaneous shock" and "structural transformation" across both semantic information and pseudo information. The paper concludes by discussing examples of operational definitions for social media and an integrative perspective spanning AI systems, nervous systems, and societies.