Linear Regression "of y on x" vs "of x on y"

Regression of y on x -> Finding the line of best fit with minimum total distance² in y Regression of x on y -> Finding the line of best fit with minimum total distance² in x How are they different from one another?