Hi ssjenforcer,
You have found perfectly right that mixing separate pure red, green and blue channels result in a white color because these 3 subsurface channels are combined using weighted sum. You can make your weights arbitrary, but they will rescale accordingly. There are three of these layers just for flexibility, so that you could make one layer thin and and another layer thick changing the radius value accordingly. For example for skin surfaces it's nice to make one red layer with larger radius to get this subtle red lighting effect when the light is behind the skin object.
Actually this SSS is like an extented Diffuse. For example you can make SSS with red/green/blue or white/white/white colors and zero radiuses and compare it with white Diffuse. They will be the same. Later you can change the colors/radiuses and weights and make a different look.
Usually SSS is combined with reflection layers to make a tint layer. But combining with Transmission or Translucent is also possible.

In the future I will make a so called random walk SSS solver to get more precise SSS for small thin object parts with large SSS radius.