Simply follow the on-screen instructions.
This was an experiment to see if you could detect whether the user was on a touchscreen reliably based on mouse movements. I believe this is also similar to how reCAPTCHAs work. There's quite a bit of math, but the important parts is basically, each time the mouse is moved, adjust the average pixels per movement, or ppm. (This is not exactly the pixels per movement due to not using the Pythagorean theorem and simply recording the changes in x and y coordinates added together.) It's pretty hard to get to above 50 average ppm using a mouse normally, but if you shake the mouse back and forth quickly you can get it up pretty high. On a touchscreen, the average ppm is usually in the 300s being that on a touchscreen, it is not one smooth movement between buttons but a jump. Basically, it works pretty well unless you try to cheat the system.
