Since I made a post in this beta thread earlier and got drawn into it, I thought it would be interesting to test the latest version (d12e) as the Soviet Union. In 1.6, the Soviet Union was vastly overpowered, but with the German AI improvements it was not clear how much of a challenge it would be for a player. I saved a game right after the German declared war in case someone wants ot check for themselves.
https://drive.google.com/open?id=1daWS-nnLv_TWxJZkan-hidvxbQIYDIEq
My overall impression was that by 1941 the Soviet Union can still field a more powerful army than Germany, have all the best techs, and a good set of leaders. However, I followed a slightly different path here than the AI does, which makes the Soviets much stronger. I will provide a some details below.
In terms of actions, the non-standard choices were the following:
1. Sending armor instead of mountaineers to Spain, which ensured a quick and easy Republican victory.
2. Supporting Nat. China with 40k old rifles and 7 volunteer divisions to slow down Japan so that China did not surrender before the US entered the war as they usually do.
4. Puppeting Finland (the AI does this too, though).
3. Giving the UK fuel as lend-lease (since the US doesn't) and air volunteers. The latter seemed a little buggy, since in contrast to the volunteers in China they were not sent home when the war with Finland started (does this not work correctly with only air wings?).
The two keys to a strong Soviet Army is 40-width infantry divisions and good top leadership. Thus after gaining volunteer experience, as Field Marshalls Zhukov and Konev could between them lead the 240 infantry divisions which constituted the bulk of the army (each with a lvl 3 General). Each division was composed as following
20 infantry, 1 marine, 2 artillery, support artillery, support AA, engineers, and recon.
Doctrine research was timed so that the width reduction was finished in early 1941. The marines and engineers counter the terrain penalties of the artillery for river crossings (and marshes).
Supplementary units included 24 cavalry divisions, 9 light armor divisions, and 1 motorized division, with medium armor divisions (1941 tanks) in training.
All artillery was 1939 models. Small arms were about 50-50 '39/'36 split, with the old models phasing out quickly into the reserve.
Overall, this force of almost 6 million was stronger than all the western axis powers put together. It seems the Soviet AI is not pursuing such an aggressive buildup (and that is probably a good thing), but for a player the Soviet Union is still a little economically overpowered.
In terms of AI behavior, two things stood out. First, in terms of the axis buildup on the Eastern Front, it looked very much the same way as it always used to, with the bulk of the forces in Romania. This gamble pays off if the Soviets fail to defend the south, but also leaves Germany vulnerable to a quick invasion by a superior Soviet force advancing along the Baltic. Once the front in the north breaks down, units are shifted north, causing a wide collapse. Even with a less aggressive strategy, a superior Soviet force that defends the south can bleed out the axis without giving too much ground. However, what paradox seems to be trying to accomplish is to recreate the historical situation with an initial Soviet disaster, followed by a recovery, and then the "steam roller." To make this happen is, of course, much more challenging than just defeating the axis, so I am not suggesting that a strong initial Soviet setup is the answer. But a more balanced axis offensive with more forces in the north could perhaps alleviate the problem of a Soviet front collapse in the south while leaving the axis AI less vulnerable to a human Soviet player.
The other comment is with respect to China. Once the Japanese naval invasions start, China is defeated very quickly. One reason is that by this time China has a large deficit of equipment due to very aggressive early behavior. But the main issue seems to be that they are completely incapable of dealing with the naval invasions. The key areas are not well garrisoned and while forces are dispatched, by the time they arrive the invading forces have already a strong foothold. But then, when either an offensive is launched in the north or another landing takes place, all defenders suddenly disappear and the invading force simply marches inland. They eventually come back, but the Japanese capture the entire coastline with almost no resistance. A less aggressive and reactive Chinese AI would probably do much better again Japan.
