The airfoil shape is to make the wind leave the blade at a different angle than it arrived, with as little drag as practical. Computing in detail the lowest-drag shapes is a job for supercomputers, so it was done by trial-and-error in wind tunnels before such computers were available. But understanding the basics is doable:
The leading edge (the one that hits you if you put your hand into a spinning rotor) should be rounded and a bit upwind of the trailing edge. The rounding lets the airflow of the "apparent wind" (the vector sum of the actual wind and the wind from the blade's motion) attach to the blade from a range of angles (for which you pay a bit of drag.)
The trailing edge should be sharp. This lets the airflow around both sides of the blade recombine smoothly and leave in a well-defined direction and without turbulence.
The hump is on the downwind side of the blade. The air is easy to control on the upwind side because if it didn't take the turn it would bump into the blade. So the blade just has to be reasonably smooth to keep the drag down, and the curve mostly doesn't matter. But on the downwind side the air has to be sweet-talked into staying with the blade as it curves from one direction of flow to another, rather than "detaching" and going off on its own, causing the blade to lose power ("stall"). Thus the smooth hump.
The blade has "twist" because the real wind is the same regardless of radius, but the wind from the motion is proportional to radius. So the slope of the chord (line from the middle of the leading edge to the middle of the trailing edge) doubles every time the radius halves. And the ideal blade width gets wider as the total apparent wind goes down. At the center this would make the blade ridiculously (infinitely!) wide and sloped parallel to the axis. But the power collected by a patch of the blade is proportional to the area "swept". So the closer to the center, the less power you lose by being non-ideal. The outer half of the blade collects 75% of the power. The outer 3/4 collects 93.75%. So the innermost quarter of the blade is pretty much just there for support and only curved to avoid drag. You can make it non-ideal, or just flat, and hardly notice the lost power. (Though its steep slope and long chord length is useful for getting the mill started up, when the blade isn't yet moving fast enough for the airflow to attach to the blades' back.)
The idea is, when the shaft is properly loaded, to make the air leave the mill with about a third of the downwind velocity that it had when approaching it. At this point the blade will be spinning with about half the RPM it would have if there were no load on the shaft and the air was leaving about as fast and in about the same direction as it arrived.
Does that help?
