Sublinear vs Superlinear Scaling

Sublinear scaling means an outcome increases less than proportionally with size; superlinear scaling means it increases more than proportionally. In a power model with positive growth, these correspond to exponents between zero and one, and above one, respectively.

For a power relationship y = Cx^b with positive size x, an exponent between zero and one means y grows less than proportionally; an exponent above one means it grows more than proportionally. Dividing by x reveals the per-unit implication: y/x falls in the first case and rises in the second.

Whether that is desirable depends on what y measures. Less infrastructure cost per resident can be useful, while more congestion per resident may be harmful; neither exponent makes bigger intrinsically better. Estimated scaling can also reflect boundaries, composition, or omitted variables. A cross-sectional relationship between cities does not automatically predict how one city will change as it grows.

When to use it

When comparing how cost, capacity, infrastructure, or output changes with system size and proportional extrapolation may be misleading.

How it can help

Derive the per-unit implication, examine mechanisms and definitions, and evaluate whether the outcome being scaled is a benefit or a cost.