TomD - I interpret what you're saying as decoupling the crossover from its environment via optimal tuning, such as the various mass/stiffness springs tend to quiet themselves quickly. That can work, but requires ongoing vigilance since the stiffness of the xo panel changes (creeps) over time. I realize that you understand the equation in other terms - transfer of energy and mechanical grounding. I don't have my mind around that, although I believe you may be onto something - your results are good.
We both agree that the xo functions better out of the box, and we understand that distance from EMF and reduction of mechanical vibration are beneficial - much more so than might be intuitively obvious. When the XO is in the box there are many assaults: the air is in episodic resonance and fluctuating pressure; the cabinet walls, the crossover panel, and the components themselves all vibrate causing microphonics and motor-currents via motion in the EMF. All considered our best production results (in the old days) were to mount the panel tightly to the cabinet and all the components tightly to the panel. The wires hung in free air, generally away from each other. In practice we controlled buzzing and rattling and considered the job accomplished when quiet was attained.
I have tried a different approach with more satisfactory results (although less good than the outboard XO.) The new approach is to isolate the various resonances, which spreads them out to various frequencies and reduces their resonance magnitudes because each individual part has far less mass than the combined whole.
Each capacitor or resistor is mounted on blue tack or mortite, held in place via its lead wires. Coils are mounted via zip ties against rubber donut standoffs. The driver leads are more carefully routed away from coils and seated in gooey tape to the cabinet walls and routed radially to the driver, avoiding the central flux axis.
The crossover is broken into individual panels for each driver, and those individual XO panels are separated from each other as far as possible and as far from the driver magnets as possible, and mounted via fiber bolts isolated from the cabinet wall via rubber stand-offs and from the XO panel via rubber standoffs. These changes would not be very expensive in new construction - they're a bit more of a hassle in retrofit, but still far less cost than swapping out components. This description is of a level 1 upgrade, probably including replacement of any and all electrolytic caps with ERSE-level PPs, and some or all resistors with Mills.
