That's actually an interesting phenomena or question. In essence sanding by hand is like a farmer plowing a field. The DA is like he plowed it crisscross and then at a diagonal angles too. So less of a grove is left showing. The effect of the abrasion is less noticeable because the paths between the grit is destroyed. But if you sanded an area by hand and sanded in 6 or 8 directions or more, you would achieve pretty much the same thing but the "random" of a ROS is designed to do exactly that.

So when you compare the result of hand sanding vs. ROS sanding, you would have to hand sand in a multitude of directions for it to be an equal test. Basically, handing sanding in one direction is going to produce larger grooves than sanding in many directions.