Dissipative Computing: A General Thermodynamic Framework for Information Processing in Open Systems

I introduce a general theory of dissipative computing—a paradigm that recognizes com- putation as an inherently thermodynamic process emerging from dimensional coherence and entropy export in open systems. Building on recent work connecting information storage- transmission trade-offs to the emergence of power-law distributions, I establish that com- putation in physical systems fundamentally depends on leveraging both threshold-crossing events and sub-threshold dynamics within a structured entropy export framework. While conventional computing treats dissipation as a limitation to minimize, this approach reveals how controlled dissipation enables computational capabilities through the formation and degradation of coherent states. I formalize basic computational operations as coherence transformations with predictable entropy costs, develop quantitative relationships between computational power and dissipative dynamics, and establish design principles for hierarchical architectures that exploit these properties. Specifically, I demonstrate that state- of-the-art AI architectures like transformers inherently exploit these dissipative principles, revealing their fundamental connection to biological cognition and physical systems. This framework unifies observations across biological, neural, social, and physical systems, revealing them as manifestations of the same underlying computational process. By grounding computation in thermodynamic first principles and explicitly addressing the computational role of sub-threshold interactions, I provide a foundation for understanding both natural and engineered computational systems as dissipative processes that leverage energy gradients to create, transform, and transmit information.