Random walks and diffusion

Take a step in a random direction, then another, with each direction chosen independently of the last. Where do you end up? On average, exactly where you started — for every walk that drifts left there is an equally likely mirror image drifting right, so the mean displacement is zero. That answer is true and useless, which is why the quantity worth tracking is the root-mean-square displacement instead: square the displacement first, so left and right no longer cancel, average, then take the root. Because the steps are independent, the squares add, and the typical distance from the start grows as the square root of the number of steps.

Diffusion is that statement applied to molecules. A molecule in air travels a fraction of a micrometre between collisions, and each collision randomises its direction, so its progress is a random walk with an enormous number of very small steps. The result is that the spread of a diffusing substance grows as the square root of elapsed time, with a diffusion coefficient setting the scale — about 2 × 10⁻⁹ m²/s for a small molecule in water and roughly ten thousand times larger in air. The square root is what makes diffusion feel so strange dimensionally: doubling the distance costs four times the time, and going ten times further costs a hundred times longer.

That scaling is why diffusion is the dominant transport mechanism inside a cell and hopeless outside one. Over a micrometre it takes milliseconds; over a metre, the same process in water takes years. Biology arranges itself around this — cells are small, lungs subdivide into hundreds of millions of tiny sacs, and anything that has to move material further uses bulk flow instead. The same square-root signature identifies diffusive behaviour elsewhere: heat spreading through a solid, momentum spreading through a viscous fluid, and the growth of uncertainty in a stock price all follow it, which is the practical reason a physicist learns the random walk once and then recognises it everywhere.