That Brownian motion did not change his mind is hardly surprissing. The theory of BM it quite neat but the reason why it was important at the time has nothing to to with randomness, it was just more evidence for the existence of atoms.
The methods used by Einstein was nothing new and neither was randomness in physics. Random variables in physics has been around for a long time, kinetic gas theory (Maxwell, Boltzmann and others during the 19th century) is inherently probabilistic and so is most other theories in statistical physics; in fact EVERY theory which includes temperature is probablistic at some level (since temperature is just a variable describing the energy disitribution of an statistical ensemble of particles).
There is a reason why the equations that are used to model Browninan motion are named after Langevin, Fokker, Plack and others; Einstein wasn't really that involved.