A very interesting essay about using some types of mathematical surfaces in architecture. The architect chooses surfaces with useful properties, like being able to walk between any two points without turning upside down, and uses them as floors, walls, and ceilings of buildings.<p>One question he didn't address was why he couldn't just create arbitrary surfaces out of NURBs or some other type of spline. It seems like that would give an architect greater freedom to mold things to his liking, instead of choosing predefined functions and playing with the free parameters. Perhaps he just likes the aesthetics of the rheotomic surfaces?