The Americans had smaller mobile units - I think it was three tank and then three armored infantry - during the war as their doctrine called for a lot more ad hoc formations; they had loads of independent battalions meant to be cross-attached as needed. The doctrines we pick ought to determine how big divisions get to be, and how much space they can be packed into.
What we really should be measuring is the "weight" of these formations. Think of it as a measure of physical mass as well as space. A thirty tone medium tank is a heavy burden, but imagine sixty horses, each weighing a half-ton - that's a lot of space. Unit deployment needs to be ruled not just by whether one has the assets to move the unit, but also whether the unit is fit to survive the environment it is being moved to. The AI doesn't have a way of seeing whether or not it will mess up a logistical situation by deploying somewhere, and when it does cause a problem, the AI has nothing to tell it to call off its deployment, so it just does so anyway, choking everything up, and also dooming themselves in the process by way of the extra attrition they suffer.
Humans can adapt easily. Any changes made will soon result in a new meta for the online community to argue about the validity of. We'd have less problems if the real issues with the game's AI were tackled, and the problems are not so separate, which is part of the reason why they have not been tackled - every change in the AI requires other changes for balance and it soon becomes a real headache.
The AI rushes too much, partly because it is drawing front lines everywhere and then trying to fill all the empty spaces. That drives the AI to put out too many land units, leaving too little manpower for air or naval deployments. For example, conscription should be broken down more [minimum age, fitness levels, length of service, etc.] with more penalties to industry, and we should have something to monitor who much of our military manpower pool industry needs, so we know when increasing conscription will start to impact industry. Anything above Extensive Conscription's 5% should really begin to hurt, OR if all the logistical needs were factored in [by a pool of manpower needed behind the front lines] then 10%. Dipping into that pool should mess up logistics. Dipping into the industry pool should slow down manufacturing and reduce quality. Introducing broader conscription should reduce unit effectiveness and speed. While a human player could take not of all this and decide not to increase conscription, the AI has nothing to tell it just how much of a problem this will be. A low factor of "AI_will_do" merely means it might take longer for the decision to be made, but that decision WILL be made at some point.
The AI needs new tools to help reduce its needs, and part of these tools should be for the Faction Leader to be the boss on land, sea and air. Everyone does what the leader tells them to do - the leader assigns responsibilities, and the faction members do the jobs they are given. There should only be one front-line per faction per theater, to stop the choking up of the logistic networks, as well as to ensure that the AI knows exactly what is in theater. While the AI does give out expeditionary forces, and even accepts them, the AI still draws its own front line and sends its own units, even though some are already there with another ally. This chokes up the logistics. If the AI can't "see" that its allies do hold certain sectors, then it will try to fill those gaps itself, duplicating and even tripling efforts, all of which is unnecessary.
Some fronts are too big for one marshal to manage. A single marshal with six generals, each with the 24 division maximum, will only give you 144 divisions, and that is not enough for Russia - either invading Russia, or as Russia invading everyone else - or for the US, especially late war. Its not good enough for the masses of smaller divisions that the Chinese are supposed to be fielding [Mass Assault, remember; not MP meta] to keep the Japanese from getting too cocky, since those divisions can and will find gaps in the lines to slip through - Japan can't be strong everywhere all at once. To manage such fronts, a third level of command is probably needed, or some manner of theater system that the AI will respect - it should not try to operate outside its theater. Every army group in a theater should have an objective state to aim for, not just the most valuable VP provinces. And perhaps a list of viable strategic targets should be compiled for all the major nations, so the AI can make choices, whether its what to aim for, or what to bomb, or what to defend.
To slow down the AI - make combat more attritional and static [nearly everyone was mired in WWI thinking] - the new system of managin offensives can be utilized. Granted, the industrial powers would be able to support more, but therein a limit can be applied. For one, nations would now have to plan and accumulate materials for offensives, simulated by the decision running. They should be limited in how many they can run at a time, based on their industry and resources. Operations at sea would take some effort, as would air offensives, so land operations would not be able to be also run at full bore, all the time. There could also be a new decision introduced for the major powers [and stronger, better organized minors] to prepare for major offensives along a broad front - call it the Big Push, which should take at least six months for most to set up for [even the US as they had to ship it all before it could be used]. An air offensive would not be able to jump from one region to another, though, but as it takes time to redeploy, defensive air assets shouldn't be so easily concentrated. We are supposed to cover all the regions. The casualty rates for unescorted missions do need to be upped, especially for those using medium bombers in a strat role.
Limiting invasions could be done by requiring Command Power to be used as a part of the planning for large landings. Small landings - a single unit of 2-4 battalions - should be free, but still would need planning. Same with paradrops - small forces could drop after a brief planning period, but large drops of multiple divisions should require command power, and also assume gliders are being used, as they were almost from the outset. This could mitigate certain exploits that are popular with some people without holding them back too much.
The AI should also have some manner of target for stockpiles when planning offensives - Asking itself, Do I have the stockpiles to replace losses quickly, so I won't lose momentum? Do I have the fuel - not just "has_fuel = YES" but enough to sustain continued combined operations for months? The AI needs to be given the time to prepare before it uses its wargoal declarations, as I've seen AI Germany declaring war on the Benelux while three-quarters of its army were still redeploying from Poland; the rest were along the Maginot line. The result - Germany still beat France, but two years later. A human would never do that. A human would have had an invasion force ready to go, even if it meant less for Poland [and a longer fight there].
Overall, I think the Devs themselves need to play the game more. They need to see for themselves just how Derpy their AI can be. It would also be nice to hear some feedback from the Devs about what they may have learned form their loss in the recent Devs vs. Influencers game. But my point is that if the AI could be given some more checks on its stupidity, then a lot of the nonsence arising from what is or is not meta could be swept under the proverbial rug. Stuff like that should really only matter to the online community, and they usually mod in whatever they need to balance the game for their own needs.
What we really should be measuring is the "weight" of these formations. Think of it as a measure of physical mass as well as space. A thirty tone medium tank is a heavy burden, but imagine sixty horses, each weighing a half-ton - that's a lot of space. Unit deployment needs to be ruled not just by whether one has the assets to move the unit, but also whether the unit is fit to survive the environment it is being moved to. The AI doesn't have a way of seeing whether or not it will mess up a logistical situation by deploying somewhere, and when it does cause a problem, the AI has nothing to tell it to call off its deployment, so it just does so anyway, choking everything up, and also dooming themselves in the process by way of the extra attrition they suffer.
Humans can adapt easily. Any changes made will soon result in a new meta for the online community to argue about the validity of. We'd have less problems if the real issues with the game's AI were tackled, and the problems are not so separate, which is part of the reason why they have not been tackled - every change in the AI requires other changes for balance and it soon becomes a real headache.
The AI rushes too much, partly because it is drawing front lines everywhere and then trying to fill all the empty spaces. That drives the AI to put out too many land units, leaving too little manpower for air or naval deployments. For example, conscription should be broken down more [minimum age, fitness levels, length of service, etc.] with more penalties to industry, and we should have something to monitor who much of our military manpower pool industry needs, so we know when increasing conscription will start to impact industry. Anything above Extensive Conscription's 5% should really begin to hurt, OR if all the logistical needs were factored in [by a pool of manpower needed behind the front lines] then 10%. Dipping into that pool should mess up logistics. Dipping into the industry pool should slow down manufacturing and reduce quality. Introducing broader conscription should reduce unit effectiveness and speed. While a human player could take not of all this and decide not to increase conscription, the AI has nothing to tell it just how much of a problem this will be. A low factor of "AI_will_do" merely means it might take longer for the decision to be made, but that decision WILL be made at some point.
The AI needs new tools to help reduce its needs, and part of these tools should be for the Faction Leader to be the boss on land, sea and air. Everyone does what the leader tells them to do - the leader assigns responsibilities, and the faction members do the jobs they are given. There should only be one front-line per faction per theater, to stop the choking up of the logistic networks, as well as to ensure that the AI knows exactly what is in theater. While the AI does give out expeditionary forces, and even accepts them, the AI still draws its own front line and sends its own units, even though some are already there with another ally. This chokes up the logistics. If the AI can't "see" that its allies do hold certain sectors, then it will try to fill those gaps itself, duplicating and even tripling efforts, all of which is unnecessary.
Some fronts are too big for one marshal to manage. A single marshal with six generals, each with the 24 division maximum, will only give you 144 divisions, and that is not enough for Russia - either invading Russia, or as Russia invading everyone else - or for the US, especially late war. Its not good enough for the masses of smaller divisions that the Chinese are supposed to be fielding [Mass Assault, remember; not MP meta] to keep the Japanese from getting too cocky, since those divisions can and will find gaps in the lines to slip through - Japan can't be strong everywhere all at once. To manage such fronts, a third level of command is probably needed, or some manner of theater system that the AI will respect - it should not try to operate outside its theater. Every army group in a theater should have an objective state to aim for, not just the most valuable VP provinces. And perhaps a list of viable strategic targets should be compiled for all the major nations, so the AI can make choices, whether its what to aim for, or what to bomb, or what to defend.
To slow down the AI - make combat more attritional and static [nearly everyone was mired in WWI thinking] - the new system of managin offensives can be utilized. Granted, the industrial powers would be able to support more, but therein a limit can be applied. For one, nations would now have to plan and accumulate materials for offensives, simulated by the decision running. They should be limited in how many they can run at a time, based on their industry and resources. Operations at sea would take some effort, as would air offensives, so land operations would not be able to be also run at full bore, all the time. There could also be a new decision introduced for the major powers [and stronger, better organized minors] to prepare for major offensives along a broad front - call it the Big Push, which should take at least six months for most to set up for [even the US as they had to ship it all before it could be used]. An air offensive would not be able to jump from one region to another, though, but as it takes time to redeploy, defensive air assets shouldn't be so easily concentrated. We are supposed to cover all the regions. The casualty rates for unescorted missions do need to be upped, especially for those using medium bombers in a strat role.
Limiting invasions could be done by requiring Command Power to be used as a part of the planning for large landings. Small landings - a single unit of 2-4 battalions - should be free, but still would need planning. Same with paradrops - small forces could drop after a brief planning period, but large drops of multiple divisions should require command power, and also assume gliders are being used, as they were almost from the outset. This could mitigate certain exploits that are popular with some people without holding them back too much.
The AI should also have some manner of target for stockpiles when planning offensives - Asking itself, Do I have the stockpiles to replace losses quickly, so I won't lose momentum? Do I have the fuel - not just "has_fuel = YES" but enough to sustain continued combined operations for months? The AI needs to be given the time to prepare before it uses its wargoal declarations, as I've seen AI Germany declaring war on the Benelux while three-quarters of its army were still redeploying from Poland; the rest were along the Maginot line. The result - Germany still beat France, but two years later. A human would never do that. A human would have had an invasion force ready to go, even if it meant less for Poland [and a longer fight there].
Overall, I think the Devs themselves need to play the game more. They need to see for themselves just how Derpy their AI can be. It would also be nice to hear some feedback from the Devs about what they may have learned form their loss in the recent Devs vs. Influencers game. But my point is that if the AI could be given some more checks on its stupidity, then a lot of the nonsence arising from what is or is not meta could be swept under the proverbial rug. Stuff like that should really only matter to the online community, and they usually mod in whatever they need to balance the game for their own needs.
- 1
