Pekka, I do agree that it will break down at the most coarse-grained level — as you say the handful of oceans and continents do create that last bit of determinism that cannot be modeled stochastically.
So there is bad news and there is good news. The bad news is that the probabilistic view breaks down, but the good news is that there are not many oceans to deal with.
Here is an old but interesting article on modeling the ocean’s thermocline at a gross level.
http://www-pord.ucsd.edu/~rsalmon/salmon.1982b.pdf
I think this was published after Paltridge’s work but he does not reference Paltridge.