I like to think about manufacturing in terms of entropy. Perhaps it is my training as a chemist, but there is a certain amount of disorder in any industrial system that has to be constantly managed. Every workpiece has a set of acceptable microstates it must occupy on the way from raw material to finished product, and while there is no conservation law for precision, precision is something you can absolutely lose whenever constraints on the workpiece are relaxed.
Entropy, in statistical mechanics, describes systems with large numbers of accessible states and high uncertainty about which state they occupy. A free rigid part in space has six degrees of freedom, three translations and three rotations. For N such parts, the accessible configuration space is 6N-dimensional.
Consider a sheet of laser-cut parts still tabbed to the sheet. Each tab constrains five of the six degrees of freedom, so for N parts you have collapsed the accessible configuration space from 6N dimensions down to roughly N. That is a dramatic entropy reduction, effectively for free. You might recognize this as GD&T's concept of a datum, but I think the argument is deeper than that. The moment you snap those tabs, entropy jumps back: every part can move in all six directions and settle into any number of tangled or interlocked states with other parts in the pile, depending on geometry.
Pick-and-place systems are my favorite example of cheaply reintroducing constraint. A robot grabbing a free part and placing it at ±0.05mm position and ±0.1° rotation drives each part back into a tiny volume of configuration space, recovering precision without much added complexity.
I am curious whether this lens of thermodynamics can be pushed further in industrial systems, whether it might produce useful policy decisions or reveal deeper truths, much like how physics principles have been applied to financial markets to great effect.