Steven Strogatz, a mathematician known for work on small-world networks and popular science books, has released a new volume titled *Big Math*. The book examines mathematical problems that outstrip ordinary human reasoning, especially those that rely on massive computational resources. It frames the rise of artificial-intelligence systems as potential oracles that could reshape how mathematicians approach such large-scale challenges.
Strogatz defines “big math” as any problem whose scale exceeds what a person can manipulate mentally. He emphasizes that linear equations and sparse matrices allow such problems to be expanded to enormous dimensions while remaining solvable. By contrast, highly non-linear systems quickly become intractable for pure analytical methods, forcing reliance on numerical simulation. The author argues that this distinction marks the frontier where rigorous theorems give way to computational experimentation.
In the book Strogatz notes that sheer size does not automatically make a problem difficult. He points out that calculus, which deals with infinite processes, fits comfortably in human intuition, whereas a model containing billions of parameters can be far less manageable. The key, he says, lies in whether the underlying structure is linear and sparse; those properties keep the computation tractable even when the variable count reaches into the trillions.
Strogatz uses partisan redistricting to illustrate big-math applications. He describes how mathematicians generate thousands of random district maps that obey legal constraints such as compactness, contiguity and equal population. By comparing an actual proposal to this ensemble, analysts can determine whether the plan is an extreme outlier, suggesting intentional manipulation. The technique helped refute a claim in an Alabama voting-rights case that no map could produce two Black-majority districts, showing that many such configurations exist.
According to Strogatz, computers first served as precise calculators for engineers and scientists, akin to microscopes extending human perception. The paradigm shifted in the 1950s when machines began to explore unknown mathematical territory. He cites the Fermi-Pasta-Ulam experiment, where a team used early electronic computers to simulate a simple lattice of atoms without any expectation of the outcome, revealing unexpected quasi-periodic behavior that sparked a new field of nonlinear dynamics.
Strogatz warns that treating AI as an oracle could blur the line between tool and autonomous discoverer. While he still views computers as instruments programmed to follow instructions, their ability to generate vast ensembles of solutions and to uncover patterns beyond human foresight raises questions about the future of proof and intuition. The author suggests that mathematicians will need to adapt, learning to interpret algorithmic output while preserving rigorous standards.