Very true...but I've looked far & wide for
a reason for that
specification to exist, and have found none. The only thing I can
think of is the possible flattening out of the balls causing them to
short against adjacent ones...but is that really possible? Not sure.
I suspect the spec may be due to what IBM found with SLT developement
in 1963 (which was basically BGA on a small scale) - having some
height allowed for a small moment arm, which was good for absorbing
the flexes due to thermal expansion as the chip heats up.
--
Will