I would state this differently: no single parameter can be optimized without taking the entire whole into consideration. Any architecture will always have strengths and weaknesses. Intelligently optimized, varying architectures converge but ultimately still carry their basic DNA (the architecture's key attributes - their strengths and weaknesses).
Cheers,
Thom @ Galibier Design
What about this scenario: Parameter is adding mass to (preferably the outside of) the platter. Surely the only downside is the weight (and thereby friction) on the bearing. If the bearing housing includes opposing magnets so that there is actually no weight on a thrustpad, then that parameter can be improved without any negative downside. Or is there some other thing that will to some extent mitigate the inertia value of the extra mass?