These guys got it right.

So... what about using ceramic bearings for cup-and-cone applications? Is that kosher or is it just asking to score the races?

My understanding is the latter, that the typical cup and cone bearing assemblies standard in bikes
for many years are designed with a certain useful lifespan that includes a few bearing replacement
cycles..........i.e. the races are more durable than the bearings, which are cheaper and more easily
replaced. But I have to admit I've never tried it.

I always like a good Monty Python reference.

Oh? Try this.

There is less friction with the nine balls and the cage. When a rolling ball touches the cage, it is a rolling ball against something not moving.
When a rotating ball touches a neighboring ball it is two surfaces going in opposite directions. That's worse. Bearing makers know how to make bearings, most of the time. Not always.

Interesting. I wonder how much the grease reduces that. It would seem the coefficient of friction for the unpolished cage could be as high as the highly polished counter-rotating surfaces of the balls. I'm sure the difference is minuscule, especially with appropriate lubrication. I was more interested in the potential for reducing wear due to greater load distribution over a higher number of bearings. I'm over 200#, so I have to think about these things. Going from 9 balls to 11 represents an over-20% increase in bearing surface.

it depends on if you add a grease with a P.E.D to it.

I'm sure it is minuscule. All this stuff is on the interwebz, I bet you could find analysis of the differences. I got this info from a bearing company, where I used to buy bearings.

You would expect them to know, I reckon.

You guys & gals are overthinking this a little bit. Assuming the intended lubrication is present, the majority of friction will be due to sliding between the ball bearings that are taking the instantaneous load, and the races of the cone and the hub. Sliding is unavoidable because a monolithic ball bearing rotates at a given rate, while the range of line contact between the balls and the races require a variable rotation rate of the ball bearing. Of course, any non-circular irregularity in the bearings or the races will add to the friction, but that is mostly independent of the number for balls.
In the first approximation, sharing this sliding friction among more or fewer ball bearings will not appreciably change the net friction. It seems to me, that the main advantage of more ball bearings is increasing their longevity and that of the races, due to reduced cumulative fatigue damage (since the load is shared over more net contact area).

^^That.

I think I remember Oprah asking this question in her interview last night.