I think that's a corollary, not the main theory. The main theory is analogous to Les' description of synthesizer voicing. Basically, you start with a string stretched across two fixed points. You pluck that string with fingertip, fingernails, a chip of (whatever), and the string vibrates. Everything on the physical structure of the guitar itself subtracts energy from the strings' vibration in one way or another, the net result of which is, generally speaking, a filtering of the sound. So that's the starting point, the basic assumption from which everything else follows.
From that perspective, yeah, in some respects you want to "minimize the things that subtract from the sound" -- e.g., if maximizing the duration of the strings' vibration ("sustain") is a goal, which it certainly is in PRSh's worldview, you want to minimize anything that makes the strings' vibration die out more quickly. But you might also want the filtering to happen in ways that "sound good" in which case you might want to filter some things a little more strongly, i.e., increase something that subtracts from the sound in a certain way, maybe to reduce the response in some part of the frequency spectrum relative to the response in other parts for a more balanced voice.
Exactly. When I refer to subtraction, I mean as in subtractive synthesis, where frequency and formant filters, envelope filters, and resonant filters operate on the waveform.
In other words, where you DO want the effect of the filtering that takes place.