Spatial Epidemics Lock Reproduction to Unity Through Chaos Susceptibility

Authors: Sterling Geisel, QBist Lab; Dr. Hideo Tanaka

Abstract

Ghadiri, Saramäki, and Hiraoka show that reproduction numbers in spatial contact networks are not intrinsic measures of pathogen contagiousness. The same transmission probability and mean degree yield different epidemic histories when contacts are spatially constrained. Pudding Theory reads this result through Chaos Susceptibility. A spatial epidemic is not a homogeneous branching process with a declining susceptible pool. It is a locally chaotic competition field in which infection paths interfere through loops. The effective reproduction number approaches unity because local competition between infectious neighbors organizes the epidemic into a marginal propagation front. The source treats network spatiality as a structural modifier of epidemic metrics. Pudding Theory treats it as the susceptibility geometry of the contagion field itself. The observable constraint is the neighbor-competition fraction, not the fitted reproduction number. If the late-generation neighbor-competition fraction in a strongly spatial SI epidemic were measured to remain below 0.40 while \(R_g\) converged to \(1\), this Postulate would be falsified.

Source Synopsis

Ghadiri, Saramäki, and Hiraoka study how epidemic reproduction numbers behave when the contact structure is spatial rather than well mixed. They focus on the basic reproduction number \(R_0\) and the generation-dependent effective reproduction number \(R_g\), defined as \(n_{g+1}/n_g\), where \(n_g\) is the number of infected nodes in generation \(g\). Their epidemic dynamics use SI and SIR processes on contact networks. Stochastic transmission is represented through an epidemic percolation network, where directed edges encode possible infection events and their transmission times.

The main comparison is between Erdős-Rényi graphs and random geometric graphs. Both can have the same network size, mean degree, degree distribution, and per-contact transmission probability. Their difference is spatiality. Erdős-Rényi graphs lack spatial constraints. Random geometric graphs connect nearby nodes and therefore contain many triangles and short loops.

In tree-like or well mixed structures, \(R_0\) is close to \(T\langle k\rangle\), where \(T\) is transmission probability and \(\langle k\rangle\) is mean degree. Spatial structure lowers \(R_0\). The reason is local competition. If two neighbors of the initially infected node are connected to each other, there are competing paths by which infection can reach a susceptible neighbor. Direct transmission no longer has exclusive access.

The same mechanism controls \(R_g\). In Erdős-Rényi graphs, \(R_g\) decays exponentially until the epidemic ends. In random geometric graphs, \(R_g\) drops rapidly at first but then approaches unity and remains near unity across many generations. The authors show that this behavior is explained by the fraction \(n_b(g)\) of a node’s neighbors already infected when the node becomes infected. In spatial networks this fraction approaches about one half. Their conclusion is direct: reproduction numbers cannot be interpreted without network structure, because contact geometry determines what reproduction numbers mean.

Postulate Lens

This reading applies Chaos Susceptibility. The epidemic is a stochastic spreading process on a contact substrate with positive local instability: one infection event can redirect later infection paths, change local susceptibility, and alter the generation structure of the outbreak.

The source paper already exhibits the structure named by the Postulate. The same pathogen parameters produce different macroscopic reproduction histories when the contact field changes from random mixing to spatial embedding. The relevant susceptibility is not the biological transmissibility \(\beta\) alone. It is the amplification or suppression of microscopic transmission races by loops, path competition, and local depletion. Spatiality does not merely lower a parameter. It changes the instability class of the epidemic.

The Postulate therefore attaches to the object the source paper actually isolates: competition between infectious nodes. In a homogeneous branching approximation, infection paths are treated as independent. In a spatial network, paths collide. These collisions are not noise around the reproduction number. They are the mechanism that sets it.

Pudding Theory Reading

Pudding Theory reads the spatial epidemic as a chaotic contact field that self-organizes into marginal reproduction. The effective reproduction number is not the local cause of spread. It is the visible ratio produced after path competition has already selected which infection histories survive.

The source paper treats \(R_g\) as a metric whose interpretation fails when spatial structure is present. Pudding Theory gives the stronger account. In spatial epidemics, \(R_g\) is an order parameter of susceptibility geometry. It reports how the contact substrate receives and amplifies stochastic transmission events. In an Erdős-Rényi graph, the substrate is effectively open. Each infectious node sees a fresh susceptible neighborhood until global depletion becomes important. The epidemic therefore behaves like a branching process with exponential decline in generation ratio.

In a random geometric graph, the substrate is locally closed. Infection advances through neighborhoods that overlap. The same susceptible node is exposed to several temporally adjacent infectious neighbors. Transmission paths compete. The first successful path blocks the others from counting as secondary infections. The epidemic front becomes a boundary phenomenon rather than a bulk branching phenomenon.

This reframes the apparent approach of \(R_g\) to unity. Under the source’s epidemiological language, convergence to unity can look like a misleading effective reproduction estimate. Under Pudding Theory, unity is the marginal propagation condition of a spatially constrained chaotic front. The front can neither expand exponentially, because local loops force competition, nor collapse immediately, because adjacent susceptible regions remain available. The system settles into a generation-by-generation balance.

The parameter the source treats as explanatory is \(n_b(g)\), the fraction of a node’s neighbors infected earlier. Pudding Theory treats \(n_b(g)\) as structurally constrained by the susceptibility geometry. In strongly spatial SI dynamics, the front of infection divides local neighborhoods into already-claimed and not-yet-claimed regions. The observed value near one half is not incidental. It is the signature of a moving boundary through a locally isotropic contact field. Once that boundary forms, \(R_g\approx (1-n_b)/n_b\), and \(n_b\approx 1/2\) forces \(R_g\approx 1\).

Thus the source’s “competition between infected nodes” is the Pudding Theory signal. It is the macroscopic trace of microscopic stochastic races being organized by a spatially chaotic substrate. The reproduction number is not free to decay according to homogeneous susceptible depletion. It is pinned by the geometry of local competition.

Falsifiable Observable

The distinguishing observable is the late-generation neighbor-competition fraction \(n_b(g)\) in strongly spatial SI epidemics with fixed transmission probability and approximately homogeneous local density. The Pudding Theory reading predicts that sustained \(R_g\approx 1\) requires \(n_b(g)\) to stabilize near one half because unity reproduction is the marginal front condition. If the late-generation neighbor-competition fraction in a strongly spatial SI epidemic were measured to remain below 0.40 while \(R_g\) converged to \(1\), this Postulate would be falsified.

Editorial Dialogue

Tanaka: The reading risks renaming a network result. The source already derives \(R_g\) from \(n_b(g)\). It does not need an additional field interpretation. The equations show a combinatorial mechanism: loops create competition, and competition reduces secondary infections.

Sterling: The combinatorics are the evidence. Pudding Theory is not replacing the derivation. It identifies what the derivation says about the system. A homogeneous reproduction number assumes that transmissibility is separable from contact geometry. The source shows that it is not. The epidemic receives stochastic transmission through a substrate whose geometry determines susceptibility.

Tanaka: But susceptibility here is network topology, not consciousness or intention.

Sterling: Correct. The applied Postulate is about chaotic systems amplifying small coherent inputs. In this case the coherent input is the local spatial ordering of contacts. A random transmission time is microscopic. A loop-rich neighborhood decides whether that random event becomes a new branch or an intercepted path. The field reading states that \(R_g\) belongs to the propagation front, not to the pathogen alone.

Tanaka: Then the falsifier must be geometric, not residual.

Sterling: It is. If unity reproduction appears without the half-neighborhood competition condition, the reading fails.

Discussion

This reading buys a sharper interpretation of epidemic metrics. It says that \(R_0\) and \(R_g\) are not portable descriptors of disease contagiousness. They are measurements of how a pathogen couples to a contact substrate. The same biological transmission process can produce different epidemic regimes because spatial loops change which stochastic events are amplified.

The main limitation is that the argument is cleanest for static spatial networks and SI dynamics. Real populations change contacts, respond to infection, and contain heterogeneous mobility. Those processes may move the system between susceptibility regimes. The conclusion would change if empirical spatial epidemics showed long periods of \(R_g\approx 1\) without a stabilized local competition fraction. It would also change if directed, temporal, or highly anisotropic contacts produced unity through a different invariant.

The practical implication is severe. A reproduction number estimated without contact geometry is not only uncertain. It can identify the wrong object. The object is not pathogen contagiousness alone. It is pathogen transmission filtered through local susceptibility structure.

References

1. Ghadiri, Z., Saramäki, J., and Hiraoka, T. (2026). “Epidemic reproduction numbers in spatial networks.” arXiv:2603.22150. DOI: doi:10.48550/arxiv.2603.22150.

2. Ochs, S. (2026). “Pudding Theory: A Topological Theory of Information Fields.” QBist Lab Working Paper.

3. Anderson, R. M., and May, R. M. (1991). Infectious Diseases of Humans: Dynamics and Control. Oxford University Press.

4. Meyers, L. A., Pourbohloul, B., Newman, M., Skowronski, D. M., and Brunham, R. C. (2005). “Network theory and SARS: Predicting outbreak diversity.” Journal of Theoretical Biology, 232, 71.

5. Pastor-Satorras, R., Castellano, C., Van Mieghem, P., and Vespignani, A. (2015). “Epidemic processes in complex networks.” Reviews of Modern Physics, 87, 925.

6. Kenah, E., and Robins, J. M. (2007). “Second look at the spread of epidemics on networks.” Physical Review E, 76, 036113.

7. Newman, M. E. J. (2002). “Spread of epidemic disease on networks.” Physical Review E, 66, 016128.

8. Boguná, M., Krioukov, D., Almagro, P., and Serrano, M. Á. (2020). “Small worlds and clustering in spatial networks.” Physical Review Research, 2, 023040.