Hi.....I suppose the aspect of friction can be a double edged sword.........if a cage is not a friction item per se, then God didn't make small green apples.....the balls as they rotate do rub against the cage.....that is a friction fact, whereas an undersized ball between two balls does not.
Just for the example, if you have 3 balls in a line and the middle one is undersize by a smidgeon, the you have NO friction as the 3 balls can rotate in unison without rubbing against one another.
That is my opinion, and doubtless the fact that the manufacturers have gone to the expense of R&D with cages must mean something we are not to know about.
One thing that can't be overcome with smaller alternate balls is the reduced number of balls that carry the load
I don't think a linear bearing block with alternate undersize balls will ever fail in average use or with the heavier loading that some like to submit their machines to.
I don't intend to take my mill apart to get the balls out and test if they are alternate sizes or not........some things are too tedious to split hairs on, but if I were building from scratch and the balls in the bearing blocks were of the same diam then I would re-ball with alternate undersize balls as I think that it would be a better solution and cost effective.
Why would a manufacturer of a cheap ball nut have all balls of the same size when the alternate size method would work better (as in smoother) without extra cost.
More to the point, you won't get cages with rolled ball screws or no name linear bearing blocks so small alternate balls will score heavily in that direction.
I'll desist from arguing further on this one as no one has shown decisively that the alternate small ball method is a problem.
Ian.