In a random walk, you take random steps forward or backward at the flip of a coin and ask how far you are from your starting point at the end. If I take 4 steps, they *could* be all in the same direction so I could be 4 steps from the start, but that’s really unlikely. Much more likely to have about as many forward steps as backward, for a final distance 0, 1, or 2 away from the start.
As I take more steps, my most likely distance from the start increases, but not in proportion. [It turns out](https://en.m.wikipedia.org/wiki/Random_walk) that if I take N steps, my most likely distance from the start is about sqrt(N).
Thus, real-world processes that behave like random walks have deviations that increase not with time, but with the square root of time.
Latest Answers