I don't know why what you got there is different from what I did. In my game Germany sent 175 divisions to the Russian border and waited until they had full planning bonus before declaring. This even involved delaying the declaration for 5 weeks after the NF was finished.
Why dont you fix shortage calculation itself. If there were no strict priority among queues but were resources distributed evenly - or maybe only a high medium low priority would exist, then the more realistic new calculus would actually make sense and would be easier to handle as well. If there is oil shortage then it is not that it doesnt impact at all tigers but fully halts all pziv production.
203 Divisions have been allocated to the southern western front, allowing AI Germany to just march on in Northern and Central Russia.
View attachment 245083
On the far west we have 60 divisions in a single province.
If there is no code in the game to stop the AI from doing things like this, then isn't the AI war just a matter of luck in terms of whichever AI does something like this first is likely to lose?
(((((Later post
I've had a sneaking suspicion that loading from a save game causes the AI to do a re-evaluation, or causes it to lose its focus / history on it's current setup, that weakens it. Is this confirmed/known? Or is it something else? Russia seemed to be doing rather well and I did tag over to Germany before tagging back to Soviets to just see what might be going on, something I normally don't do. So is loading from a save game potentially AI breaking? Or how about tag switching with human_ai enabled? Because this is how 1941 ends:
In the post you are quoting, there was no reload. It was a straight playthrough from 1936 to 1942 using human_ai for Russia. In the second post where AI Russia invades most of Max Boosted AI Germany, I reloaded from the 1941 save. Just to be clear on the differences in the two posts.
In theory there are probably ways to fix this in defines.lua file. Some weights make the ai stick to a plan it has set in motion, probably a faulty modifier somewhere in the base code that gives that plan increasing "score" the more units are massed there. When the AI is realoaded it recalculates its scores without the weight biases to current plan (since it reloaded it has none) and it suddenly tries to fill the front again.
When I was a Test Engineer at MS then we were able to force to build in testability features.
It is not only shortages that should be spread more evenly across the queues, but the effects of bombing/sabotage as well. This would be more realistic from an historical standpoint, as well as less of a hassle for the player. I presume it would also punish the AI less severely for its decisions than does the current system.
This is a big problem right there. All game design decisions should begin and end with the AI. I rather get along without a feature than having one which the AI cannot handle. AI should not be an afterthought.
