Maybe I'm missing something here, but there doesn't seem to be anything that guarantees the same number of guests enter each maze and the longer mazes seem to be placed farther from the entrance. Wouldn't this bias the results? (Of course the conclusion that the maze gives the pathfinding algorithm a hard time is the same.)
The guest capacity of each maze is capped, and it looks like the park is being artificially stuffed with enough guests to fill each maze. The time it takes guests to percolate from the entrance to the back of the park is probably insignificant and for almost all of the experiment's run time, all mazes will be full with a standing queue.
