My understanding is that when the wood is tension strong you can trap the back, making a smaller area of wood do more work - which is then not a problem as the wood is tension strong anyway.
Quite often the wood is rather strong in tension and somewhat weak in compression. Thus you want to maximize the area of wood that is the farthest from the neutral plane and undergoes the most compression - the belly. This prevents damage on the belly that results in set (or even worse - chrysals).
One obviously doesn't want to exceed the limiting parameters of the material. The stresses are greatest on the outermost fibers of the bow. That's why it's important to keep the back in one growth ring. A flat belly just increases the area on that the stresses (not sure about the correctness of the expressions I'm using, but I guess it convey the point) can then be distributed so that the cell walls don't collapse.
Probably all of this could be calculated in the confines of the bending a slender beam equation. However, with the nature of wood it's more of a guideline, but a doubtlessly helpful one at that, as it helps with understanding the principles of bow design.
As far as I know it's not obligatory to trap the back of the bow if you keep the belly flat. As a fellow beginner I'd suggest you leave the back alone (apart from the rounded edges) and keep the belly rather flat, thus making your bow's cross section akin to a rectangle. This helps with preventing set due to tillering errors that greatly compromise the belly resulting is set and finally sluggish, disappointing bows.
In general, sharp edges are not good, getting edges down to a nice radius helps a lot. Gradual transitions are very important when (if) you decide to make a bow with a stiff handle. The fades need to be gradual, in a radius.
Edit: I might be overexplainining things