42
(I love these overly broad questions!)
Since the top is more important structurally and acoustically, let's look at it first. If you assume that one objective is to make as powerful a guitar as you can ("Give 'em volume, they'll hear tone" John Calkin in 'American Lutherie') it follows that you want to make the top as light as you can, while still being stiff enough to handle the load. What's most important is resisting the torque load of the bridge: usually the tension load per se would not be a problem if you could take it on the central plane of a flat top. Running the strings inside a tunnel would make them hard to play, though.... You can see that there are several variables here; the height of the strings off the top, the span of the top between the tailblock and the waist bar (on a classical) or UTB (on a steel string), the exact material properties of the top wood, the bracing design/profiles/material, how thick the top is, the phase of the moon, the outcome of the Greek elections.... Well, you get the picture. Let's suppose you figure out some way to determine what the 'right' thickness for the top is given all the different variables (let us all know when you do, please!): what happens if you go thicker or thinner? Let's assume for the moment that you can compensate with the bracing and other design aspects, so that the top will still be stiff enough, at least in the short term.
Let's say you made the top thick enough so that it needed no bracing at all. It would hold up OK, but it would end up really heavy: the whole idea of bracing is to get added stiffness without too much weight. Since the strings don't have a lot of horsepower, you'd lose acceleration (high frequency response) and top speed (power) as a result of a poor power/weight ratio. Going the other way; using a really thin top with lots of tall, thin braces close together (to avoid buckling in between) makes a light top that's really responsive, with a lot of high end. Maybe too much. Smallman's 'lattice' braced classicals are like that, and some folks find the sound less than appealing, albeit LOUD (Calkin isn't gospel).
It's possible to make a top that is really thin, and lightly braced, that will be responsive and LOUD AS HELL, and will hold up OK - for a while. Some folks call that a 'Ferrari guitar'. Only you and the customer can decide how far you want to go in that direction. If you're going to stick with wood as a material it's not possible to make a guitar that will last forever and sound great. All guitars eventually swallow themselves through their own sound hole. Sorry.
We get into a similar situation with the back, even though that accounts for less of the sound. From what I can see, you want the back to be somewhat responsive in the low range, but less active as you go up in pitch. A really 'dead' back, either through stiffness, like on an Ovation, or by being really massive, might not be able to add to the low end power like a more flexible one can. In the higher frequency range you don't want it moving too much, taking too much power out of the top (where else is the power going to come from?), but some limited activity seems to help with 'tone color'.
Summarizing with desperate brevity, we're looking for that 'Goldilocks' spot; the 'right' mass and stiffness for the guitar. It's something that's difficult to specify exactly, given the variability of wood, and will be (much!) different for your Jumbo than it is for your buddy's Classical. An awful lot of that is taken care of in the standard designs, which have been sweated out so that they work reasonably well if you've got 'average' wood and do a decent job. Experience can help you make improvements, and there are measurements, such as 'deflection tests' and 'Chladni patterns' that can make the process more rational and short cut some of the experience. Given the inherent variability of wood, even within a given species, you can't just specify a top thickness based on the species (although there are lots of folks on these lists who will be happy to do so): you have to measure the stiffness somehow if you want the best answer (as opposed to a ballpark number).
)