I would try the two highest for the first two transistors to maximize total gain: the first two transistors are in a Darlington configuration so the total effective hfe is roughly the product of the two...
I love how heavy fuzz gets that initial attack through. Like using a gate with a slow attack to let the initial burst through. I’ve been trying to find just the right heavy fuzz that doesn’t get muddy.
Small signal model says all gains multiply. Since multiplication is commutative, order doesn't matter. Distortion circuits are all about the large signal behavior. There, we care about transfer function, headroom, overload recovery, bias shifting, that kind of stuff. Just sayin'.