Ok,
Just so I am catching what you are saying, cottonwood... The loses due to aerodynaic drag and ground friction is greater than the loses of putting energy back into the battery? I know I have seen elsewhere on this forum that coasting down a hill is better than trying to regen the battery.
As they say, it all depends.
Transfers of energy from potential to kinetic and back are lossless, and it costs about 10% in losses to put energy into the battery and another 10% to take it back out for a net 19% loss; we will round that to 20% in our gross, hand-waving level of precision. The penalty for speeding up on down hills and using that kinetic energy on up hills is that you are traveling for some period of time at a faster speed and losing extra energy to aerodynamic drag. There are many losses for the MS, but the part that goes to pushing all of that air out of the way (aerodynamic drag) scales with the square of velocity.
If you are doing a little "whoosh" through a small valley, then you are not at speed very long and the aerodynamic losses are small relative to the 20% or so losses putting energy into and then taking it out of the battery. So for small rolling hills, its better to let the car roll, convert potential energy (altitude) to kinetic energy (speed) and back.
The situation changes when you have very long descents with big altitude changes. My Eisenhower tunnel to Silverthorne, CO example was a good one. On big, long hills, the car accelerates to terminal velocity. Once at terminal velocity, all of the that great potential energy is just being converted to heat in the air as aerodynamic drag, and no more energy is being added to the kinetic bucket. Those aerodynamic losses just goes on and on...until you reach the bottom of the hill. So for the long, mountain descent case, even with the 20% round-trip loss to and fro the battery, its better to regen than modestly heating up that mountain air. So on long, big descents its better to use CC, regen, and put something into the battery.
Where is the tradeoff, break-even point? I don't know. It would be a great analysis that I may do some day.