Can't speak for @snakker, but what I understand, or rather what I do now, is that I'll add a layer, use the whole canvas space to draw, and when I'm done, hide it, add another layer, and work on the next gesture/sketch. By the end I can rearrange them as needed, making them smaller, moving then, etc.
Using the whole space allows me to do larger arm/hand movements and to visually focus more on the current gesture, which usually results in smoother lines.
What I did before, was just draw on a single layer, zoom in, and do all drawings one besides the other, trying to take up as much canvas space by the end. Some do gestures as long as there's space in their canvas.
I think either approach is viable, as long as you are comfortable with it.
