My last frustrating attempt at the game as GER, roughly 5-10 units were stuck waiting for supplies after advancing only about 2 provinces deep into Poland against light opposition, despite having been in supply for weeks or months while awaiting the order to attack.
5-10 divisions of the German army in Poland at any given time is not so bad, beating Poland is still very easy on any difficulty level. It may even be realistic for the logistics to be screwed up a bit like that.
But later on, when you hold France, half of Russia and maybe parts of Africa and the Middle East, the supply system really crumbles. Unless you keep your army unnaturally small, you risk having almost all your units out of supply, due to the bottleneck at/near the capital, combined with supply flowing back and forth like some kind of standing wave.
If all provinces that are within the same rail network are treated as ONE province for supply purposes, these fluctuations should be eliminated, and the coding of the algorithms should not be too difficult. (The GUI for the rail network needs to be added however). It should also help Japan in China, as long as they keep to the part of the part of China that had a rail network, while keeping Japan away from the inner parts of China, where rail did not exist. (To conquer China, Japan would need to extend the rail network to the interior of China.)
For overseas supply, each continuous rail networks should be treated as one "continent". If, for instance, Japan has holdings in Russia, China and India, all connected by land, but not by rail, there should be (at least) 3 different supply dumps.
Obviously, to compensate for the way supply is eased by (functional) railroad, overland supply tax and throughput could be a lot harsher than today, especially in regions with low road infrastructure and difficult terrain.
At this point, almost any change has to be an improvement.
While it can seem that way, you should not underestimate the number of non-functional supply systems that can be created