Dale, here's how I see it... wood or glass, same same. When shooting 3 under, the fulcrum point of the string hand is generally closer to bow center. In fact, depending on design, it may be exactly at bow center. If the bow hand and string hand fulcrums are at bow center, the bow will end up tillered even in order for the limbs to flex evenly... assuming they started with the same profile and let's say, both with 1" of reflex for instance. It's when we depart from this symmetry that things get harder to follow.
My split finger bows usually end up a little weaker in the bottom limb than the same bow tillered for 3 under... and I know that goes against what some recommend, but I balance them relative to my holds and that's how they end up. Folks who tiller them to a predetermined measurement and presume they'll act a certain way have yet to convince me of it, or explain why mine are often different or 'opposite' from theirs yet feel perfectly balanced in the hand, why tiller doesn't shift, and why I don't have to move the nock points to 'tune them'.
There's a little more involved than simply where the string is held, and notions like 'moving the hand down puts more stress on the lower limb' aren't telling enough of the story.
Moving the string hand influences where the dynamic fulcrum is, i.e. the bow's balance point during the draw, up to and including full draw. This balance point dictates the virtual limb lengths. For instance, if we do nothing more than move the string hand up higher on the string, the dynamic fulcrum moves too... and when it moves up, it makes the virtual bottom limb longer. Since a longer lever has a mechanical advantage, it now needs weakened in order to match the limb movement of its counterpart.