I built a test circuit with 3 pickups of 8K each having a 470K log blending pot and a 1n cap. Bridge and neck each have a phase switch. The 500K master volume is inside, while the blend pots were mounted outside the guitar to be able to crocodile some extra stuff to the lugs, or to instantly turn them into shunting pots, by shorting cw lug to wiper. Doing that and using such large pot values, the differences between shunts and dividers become more apparent.
Generally, volume control is spread more evenly over a divider's rotation, while with a shunt smaller value pots are needed for that, and the smaller they are, they eventually affect the pickup's presence, and a tradeoff between presence and even control emerges. I prefer little resonance loss for each pickup alone.
I didn't try the circuit without bypass caps, but they appear to do a good job letting signal through around the pot's resistances. Each pot, combined with it's pickup, has a series resistance of (470 + 8) / 4 = 120K at it's resistive midpoint (which is not the same as the rotation midpoint, since they are log pots). So without caps and with all pots at their resistive midpoint, each pickup signal would have to traverse an additional resistance of 240K. How much the perceived volume loss would be, I don't know, because I didn't test without caps.
That immediately leads to another difference compared to shunts : while with shunts, the combined resistance seen by the next stage can only become smaller while turning them down, the resistance with dividers goes up and then down again. At it's midpoint it totals up to 360K, and that in parallel with the master volume's resistance, is still 200K. Turning master volume down a bit, it peaks to (500 + 360) / 4 = 215K. Regarding hum, a good ground path from the wall supply is the way to stay out of trouble here, as it is with many guitars.
It's nice to have the out-of-phase sounds under control (I like to stay away from boinginess), but there are slight volume drops in the middle of the pots' travel (an exception is the bridge pot), as if the bypass caps weren't there to prevent that. When changing a pot into a shunt by shorting cw lug to wiper, it doesn't go away, it only relocates to a different pot rotation point; the same with adding resistors in parallel over the 1n caps. So I wouldn't say it's inherent to a voltage divider. I guess it's a point where the complicated dance of frequencies phasing each other out is going strongests, and that point isn't necessarily at the end of a pot's travel... Oh well, I am already glad that there is a lot of control available to get usable out-of-phase combinations. Later on I will try pots of somewhat smaller value, and possibly sub the 1n caps with caps-resistor combinations.
This is only a first impression and not a very detailed comparison to a circuit using shunts. I was only testing it for a few hours, into a Boss GT-5. I feel it's a thing you have to learn to know how to use it (I always start setting one pickup at maximum and use one or both of the others to morph a sound). The differences can be very subtle and can easily go lost with too much gain, it's more a thing for clean amp settings. But some out-of-phase combinations can work well with for instance a fuzz (but I suspect going straight into any low input impedance like a Fuzz Face pedal or even a line input will have a bad impact on the blend controls; the GT-5's input is 1M I think, no problem there).
There's no quick way to change from one sound to another (except with the phase switches), and reproducing a sound means memorizing pot combinations, so this thing has a longer learning curve than a thing with switches. My preferred Strat circuit has a bit of both and after this test, it still is.
