Diffundeerub

Diffundeerub is a theoretical construct in network science and information diffusion that describes a diffusion process in which quantities held by nodes in a network are redistributed across neighbors in a way that alternates between forward propagation and partial retraction based on local conditions. It is used primarily in conceptual models of how ideas, items, or signals spread with intermittent backward flow.

Origin and etymology: The term was coined in the early 2020s by researchers exploring diffusion with mass-balance constraints. It blends diffusion with a metaphor of rubbing a substance to encourage even distribution. The term appears across several collaborative works rather than belonging to a single inventor.

Mechanism: At each time step, each node retains a portion of its state and distributes the remainder to neighbors according to a diffusion kernel. After distribution, a rebalancing rule adjusts each node's state by a local, non-linear function that encodes fatigue or saturation. The result is a dynamic where diffusion proceeds with intermittent retraction, producing non-monotonic spread patterns.

Applications: Applications include modeling information spread in social or sensor networks, testing privacy-preserving diffusion mechanisms, and studying rumor dynamics with fatigue or scrutiny effects. It provides a framework to explore how network topology and local constraints shape the reach and persistence of signals.

Criticisms: Limitations include a lack of standardized mathematical formalism, sensitivity to parameter choices, and challenges in empirical validation. Critics argue that diffundeerub remains largely theoretical, with real-world diffusion influenced by asynchronous updates and external factors not captured by simple models.