Information Propagation
Information propagation is the process by which signals, states, or patterns of behavior spread through a network or system, altering the states of nodes as they travel. Unlike simple diffusion — which treats information as a passive substance spreading according to concentration gradients — propagation is an active process in which the structure of the network, the rules governing node behavior, and the content of the information itself co-determine the outcome. A rumor propagating through a social network is not the same phenomenon as heat diffusing through a metal plate, and the mathematics of diffusion equations will mislead you if applied uncritically to information systems.
The key insight is that information propagation is path-dependent and context-dependent. The same piece of information — a financial warning, a public health alert, a software vulnerability disclosure — will propagate differently depending on the topology of the network it enters, the trust relationships between nodes, and the prior beliefs of the receiving agents. This makes propagation a fundamentally systemic phenomenon: it cannot be understood by studying individual nodes in isolation, nor by aggregating them into a mean-field approximation that washes out network structure.
Mechanisms of Propagation
In simple contagion — the classic epidemic model — a node becomes infected (receives the information) upon contact with a single infected neighbor. The SI, SIS, and SIR models from epidemiology apply directly: the basic reproduction number R₀ = β/γ, where β is the infection rate and γ is the recovery rate, determines whether propagation dies out or achieves global saturation. These models work well for simple, binary states: a piece of information that, once heard, cannot be unheard.
But most information does not propagate like a virus. Complex contagion — introduced by Damon Centola — describes situations where adoption requires reinforcement from multiple neighbors. A person does not join a protest because one friend mentions it; they join because three friends mention it, and the cumulative social proof overcomes their inertia. Complex contagion produces fundamentally different dynamics: it spreads through clustered networks but stalls on random networks, the opposite of simple contagion. The distinction is not a minor technical refinement. It determines whether information spreads through weak ties (simple) or requires closed triads (complex), with profound implications for everything from political mobilization to technology adoption.
Propagation in Cascading Systems
Information propagation is the accelerant of cascading failure. In power grids, the failure of one transmission line changes the load distribution across the network, and the information about this change — encoded in voltage fluctuations — propagates at near-light speed. In financial markets, the information that a major institution is insolvent propagates through counterparty networks, triggering margin calls and fire sales that amplify the original shock. In supply chains, the information that a critical component is unavailable propagates backward through the network, causing upstream firms to halt production and downstream firms to accumulate inventory.
The critical feature in all these cases is that the propagation of information is faster than the system's ability to adapt. The power grid cannot reroute power as fast as the failure information travels. The financial market cannot reprice assets as fast as the solvency information spreads. The supply chain cannot adjust production schedules as fast as the shortage signal propagates. This temporal mismatch — between information propagation speed and system response speed — is the defining condition for catastrophic cascades. It is also the reason why circuit breakers, buffer stocks, and redundancy exist: they are not responses to the shock itself but responses to the speed of information about the shock.
Propagation and Emergent Computation
In systems that perform emergent computation, information propagation is not merely a side effect of network dynamics — it is the computation. In cellular automata like Conway's Game of Life, the propagation of patterns across the grid performs logical operations. In neural networks, the propagation of activation signals through weighted connections performs pattern recognition. In ant colonies, the propagation of pheromone signals through the environment performs path optimization. In all these cases, there is no central processor executing a program; the computation is the propagation, and the propagation is the computation.
This perspective reveals a deep connection between information propagation and synergetics. In Haken's framework, the slow macroscopic order parameters enslave the fast microscopic variables. The order parameter propagates its influence through the system, and this propagation is what maintains the coherent pattern. The slaving principle is, at its core, a theory of hierarchical information propagation: information about the collective state flows downward, constraining the behavior of the components. The propagation is not merely physical; it is informational, and the information being propagated is about the system's own organization.
The persistent conflation of information propagation with physical diffusion — treating tweets as molecules and retweets as collisions — is not merely a category error. It is a methodological blindness that prevents the social sciences from developing genuine theories of collective behavior. Information is not substance; it is difference that makes a difference. Until we build models that respect this distinction — models in which what propagates matters as much as how fast it propagates — we will continue to be surprised by cascades we should have anticipated and seduced by equilibria that do not exist.