What is a maze?

A maze can be formalized as a graph, but physical solvers do not receive a graph. They encounter walls, gradients, costs, and partial observations. This note uses that gap to compare classical search with embodied and nature-inspired computation.

The same distinction between a pattern and the process producing it appears in the pinecone inquiry, where local growth performs work that a global geometric description leaves unexplained.

Questions this note must answer

• Which maze-solving algorithms require a global representation?
• How do memory constraints change completeness and optimality?
• In what sense do slime molds or reaction–diffusion systems “solve” a maze?
• What do physical solvers compute for free through their dynamics?

The finished version will include visual algorithm traces and a small reproducible benchmark.