The Law of Drift: How Signals Attenuate Over Time

Across physical, computational, and institutional systems, representations degrade as they move through distance, time, and layers of transformation. This paper proposes a unifying statement: drift is information attenuation. We formalize drift as a decline in the information a representation carries about its target state, and we show how this decline follows from standard results in information theory (notably the data processing inequality) once one admits that (i) information is data under context, and (ii) real transmission and summarization pipelines contain lossy steps. We then connect the resulting law to a family of relational drift models: Structural Distance Theory (SDT), Truth Stabilization Theory (TST), Relational Systems Theory (RST), and a signal-routing model of lossy projection under pressure. The result is a category shift: drift is not primarily a sociological pathology but a structural constraint on information instantiated in physical and contextual media. Stability is therefore an engineered achievement requiring ongoing maintenance (repeaters, error-correction, audits, dissent channels, instrumentation, and other corrigibility scaffolds).