"....the current will flow into the properly placed resonant radials without insulating anything."
no it doesn't. if this was the case then everyone who simply added a radial system would report improvements in performance. this forum is replete with posts reporting exactly the opposite scenario. simple rf current measurements comparing the total current in all radials to the current in the radiator will bear this out. until total radial current equals radiator current the system is operating out of balance. furthermore, a 1:1 current balun is (required) not an option as long as unbalanced feedline is being used to feed a balanced antenna design.
the best materials i have found for isolating the support structure from the antenna are teflon and delrin because of their high (k) dielectric properties. the objective is to insure that all common mode current flows ONLY in the radiator and the radial system, thereby insuring that maximum current in the e-plane is transferred to the initial primary electric field. this condition is defined as "perfect elevated ground".
reference:
E-PLANE H-Plane
as was previously posted, a reduction in overall average noise in the system is the result of the suppression of common mode current present on both the support structure and the outer shield of the feedline.
signal strength increases on the order of 1.0 - 2.5 S-Units are not actually occurring as the common s-meter (asynchronous) reads much less than 6 dB. per S-Unit below S-9 but improvement in signal strength does occur nonetheless. my measurements indicate that actual figures are more on the order of 1.5 - 3.75 dB..