I hadn't thought of it that way, but it's true. The motion of the flakes is ultimately determined by the sum of the impulses it receives from the molecules hitting it, along with a general downward drift. So I went out to track the micro wind patterns in a cubic meter of air. I started by tracking the molecules in the air and counting the individual hits as they bounced from flake to flake. This proved difficult and I soon tired of it. Then I reverted to watching the flakes themselves. Unfortunately, just as I got caught up in the wind-tossed life of a flake, a gust would take it away from me and I was overwhelmed by the sense of loss. I would focus on another, but you know how it is with snowflakes. Next, I gedanked about the following experiment. Place two film (not video) cameras at right angles to each other so that they cover the same volume of air. Their shutters should be synchronized somehow so that each frame from one camera can be matched to a frame from the other. Film for about a minute. Then see if you can identify the individual flakes from each camera. If you can, you will no longer need a weatherman to know which way the wind blows.I think it's generally accepted as Brownian motion with drift.