Looks to me like that winding puts the current through more copper than necessary, resulting in higher I2R losses.
Another way to look at is is that you're summing two voltages at a 30 degree phase angle. Assuming a sine wave you'd be losing about 7% of the voltage (and thus the power at a given current) that you could have gotten from the same amount of copper and magnets in the same coil arrangement, but with twice as many turns of half-the-cross-section wire, connected in a 6-phase system. (The actual amount would depend on the waveshape, of course. Perhaps this arrangement would make for a sufficiently flat-topped waveform that you'd do as well, or even better than, the more traditional setup. But at first cut I'm inclined to doubt it.)
Regardless, I'd expect that adding more magnets in the more traditional style would have more copper under poles, raising the voltage and improving the ratio of power out to stator heating.
I'd love to hear your EE friend describe the logic behind his recommendations. Could you get him to compose an explanation of what he was trying for?
