My limited experience also shows that thinner glass (given good core wood) makes a smoother bow that is soft in the hand. But there starts to be a trade-off in performance.
As for why not build an all glass bow? In a bending limb, the outer fibers take the most stress and strain. You have tension on the back and compression on the belly. The middle (or neutral axis) is where the tension and compression meet and the net strain and axial sterss is zero. What you need in the middle of an ideal composite laminate is a core material whose only job is to keep the outer fibers separated. If you make an all glass bow, you're using very heavy glass with good axial strain characteristics where it has no benefit. That weight will hurt your limb performance.
An ideal laminate (which you can see on ILF olympic limbs, high performance sailing, etc) would be outer fibers as thin as possible with maximum axial strain carrying characteristics -- i.e. carbon fiber. The core will be light and have no axial properties to speak of but will hold up very well in shear (the force of the two outers skins trying to pull together as you bend the limb) -- i.e. foam or end grain balsa.
For those of us making wood cored limbs, the wood laminates do carry some of the strain energy, even though the glass carries more. As a result, there is a point at which too much glass makes the limb heavy and slow and too much wood makes the limb less snappy. For a given limb design, this optimum balance could be calculated using Young's modulus and a bunch of dynamic finite element calculations for all materials and masses involved. More than likely people wanted to avoid that kind of migrain and just went with trial and error to find that 20-25% was generally close to the optimum point.