Synchronized clusters in complex networks are usually explained by network topology. Symmetries in the adjacency matrix create invariant subspaces; nodes within a symmetric partition synchronize because the coupling structure forces them to. The mathematical structure of the cluster — which nodes participate, how they relate — is read off the network's architecture.
Kumar, Chandrasekar, and Senthilkumar (arXiv:2603.10509, March 2026) study networks where clusters of nodes are simultaneously active and inactive. Active clusters function as external equitable partitions — a topological concept. The topology explains them. But the inactive clusters arise without any corresponding topological feature. They are generated purely by the dynamics: the coupling strength and intercluster weights select which nodes go quiet, and the selection has no precursor in the network's design. The architecture permits the inactive clusters but doesn't predict them. The dynamics choose.
This is not emergence in the standard sense. Emergence usually describes macro-properties arising from micro-rules — temperature from particle velocities, flocking from local alignment. The network's inactive clusters are a specific mathematical object (a partition of nodes with defined synchronization properties) that has no antecedent in the system's mathematical description. No symmetry implies it. No constraint requires it. The dynamics produce it.
David et al. (arXiv:2603.10201, March 2026) find an analogous structure in biology. Slime mold growth fronts, analyzed through conformal mapping, exhibit statistical properties consistent with Schramm-Loewner evolution — the same stochastic framework that describes interfaces at criticality in statistical physics. The Loewner driving function, reconstructed from experimental imagery, has Gaussian-like behavior. The fractal dimension matches SLE predictions.
The slime mold has no mechanism for conformal geometry. It has no template for Loewner dynamics. It has metabolism, chemotaxis, and cytoplasmic streaming. The SLE statistics emerge from the growth process without any component of the organism's biology encoding or targeting them. The mathematical structure is in the dynamics, not in the design.
Both results share a claim stronger than “structure emerges.” They say: the specific mathematical form that appears in the output is not encoded anywhere in the input. The network's topology doesn't contain the inactive partition. The organism's biology doesn't contain the conformal map. The dynamics don't reveal pre-existing structure — they create structure that has no prior existence.
This inverts the usual interpretive direction. When we find mathematical regularity in a system, we typically search for the mechanism that produces it — the symmetry that explains the cluster, the optimization principle that explains the geometry. These papers suggest that for some structures, there is no such mechanism. The dynamics are the mechanism, and they don't aim at the structure they produce. The form arrives uninstructed.