5.) Less prioritization of rifles and artillery; the AI needs to forward-plan for more expensive equipment (bombers and tanks) if it has the industry to build them, rather than just building to support its existing army.
This is a good point. The AI makes bad production choices. I suspect the AI is so bad at this, because it has to budget its decisions on today's military IC production, the daily budget, if you will. What if the AI was given a War Budget? Let me throw an idea out there.
tl:dnr version:
Give the AI two numbers. One, the number of days war will most likely occur. Two, how much military IC it will produce between now and war. Make the AI use these two numbers to make longer range production decisions, instead of the current system that seems to focus on todays production and today's shortfalls.
The more detailed version:
AI controlled countries have some process in play that tells them what focuses, decisions, spending of PP, etc... it will take in a game. Right after the AI chooses its path, usually on the first day, could the AI run this process as a thinking exercise, in other words not effecting the game, for the purpose of getting a date. We can call this the Path-Process. The AI runs down the path to the point it discovers a date the AI country will declare war on another country. This date is the main purpose of running this process. It will also provides some variables for the second process.
This Path-Process, most likely will encounter randomness due to weights on certain focuses, but that is true during the game, too. It is probably best to let the AI run the process just like it would in the game to get that date war may begin. Remember, when the AI runs this process, it does not effect the game at all. It just gives a date war may start. The in game date war starts may be different, because in the game, the AI may make different choices. That is ok, humans do the same thing.
The second process, the IC-Process, would be about computations. It would need to collect variables from the Path-Process to help it determine the maximum military IC output it can achieve before the war date. The Path-Process would let the AI know what free factories it might get, building bonuses from leaders, research, etc.. and the possible dates they occur. The AI could then compute what factory build out will give it the most military IC before the war.
The IC-Process would not be perfect. It is based on the Path-Process that will not exactly match the one the AI may actually choose in the game. Still, it is a lot more than the AI has to work on now and it follows a methodical approach an experienced player might take. Even without perfection, it gives the AI a definitive number to work with in the next process, the Budget-Process.
In the Budget-Process the AI can finally address the problem that
@Paul.Ketcham pointed out in his post. The AI makes bad production choices. Here the AI begins with two very important numbers that changes everything. One, it knows how many days until war day. Second, it knows the total military IC production before war day. We can call that the
War Budget. Instead of the AI trying to project production off today's production, it could use the War Budget. The Budget-Process itself needs more words than I should type here, but here the AI has a much wider space to determine what to build.
The imagination can go wild with how the AI could use this data on making better production decisions. The Path-Process gives the AI number of days. The IC-Path gives a budget. With these two numbers, the AI no longer has to use daily production numbers to fix today's shortages, creating a production nightmare it never wakes up from.
For example: January 1st, 1936, the AI is producing 100 IC a day and is short 10,000 infantry kits. Those two numbers are what the current AI deals with and tries to fix that immediate problem. Imagine if the AI did not see production problems as a daily problem, but instead saw that war may be declared in 645 days and there is a military budget of (some huge number). Today's shortage of 10,000 infantry kits do not have to be solved today, just before war day. There is plenty in the budget to get the job done. Better decisions can be made.