https://github.com/Kondikka/expertai

Summary on mod-stuff

This update focuses on optimizing the AI's game plan more so than implementing economy side systems to help me do so, as the last major update did. Obviously there's the rework of Germany to deal with, but USSR and Japan had some major weaknesses with the type of army they built, and some general research and construction priorities needed optimizing.

The second focus of this update is on the waging war part. The scriptable AIFC strategies open up a whole another dimension to try out some fun stuff (for my own amusement of watching the AI go, whether you like it or not and regardless of how much it improves things).

The AI should generally have more armor now that it can put them to a useful purpose. Along with reworking regular AI strategy scripting for smaller scenarios, in this update the mod implements several new specialized scripted military strategy systems (Barbarossa, Pacific War, Allied/Axis invasions) that are intended to help the AI plan offensives on a larger scale, that combine all scriptable aspects of AI military under a larger scope decision making process for these specific fronts and operations.

Regarding the AI for the new DLC mechanics - raids and special projects - they are essentially not modified by EAI, nor the new unit types implemented by EAI. If something turns out to be very powerful (yet somewhat historical) that should be taken advantage of, I'll look into it later when I have a better picture of how things stand in the game.

Is the new DLC required? Technically no, the mod should work regardless. Is the mod tested and balanced with this DLC enabled? Yes. Could there be a bad effect on EAI when playing without the DLC? I think it is unlikely.

Settings

* Added an eastern front EAI setting: historical. This is for when you really want to make the eastern front progress historically between the AIs, or at least make it less likely to conclude prematurely. For instance, if you want to give Germany tons of buffs but don't want it to demolish AI USSR or vice versa. What it does is it makes the AI more passive in its attacks if it controls more of the eastern front than during that phase of the war historically.
* Removed a number of EAI general settings that are outdated or relevant anymore

Fronts

* Reworked most front related scripts in the mod to take advantage of the new AI force concentration scriptable strategies: fronts in Europe, Russia, China (to be clear: the vanilla game has scripts for a lot of these cases but this being an AI overhaul mod has its own way of scripting things, and these scripts have been reworked.

Barbarossa

* Something experimental for the eastern front using the new AIFC scripts: to simulate the historical pushes (towards Moscow, Leningrad, Stalingrad and vice versa) as well as to make things more difficult (in theory): the AI divides the eastern front into sectors and tracks how these sectors progress over time. Using this data, the AI selects a sector to focus on with its AIFCs. This system is *intended* to guide the AI to exploit breakthroughs or counter breakthroughs in fast moving fronts, prioritize dangerously lopsided areas, to try pushing through a different area of the front if it is stuck.

Allied invasions

* The four major invasion operations - Torch, Avalanche, Overlord and Dragoon - are conducted under a specially tailored AI system (not the other system this mod already had). As the historical date of these invasions approaches, the likelihood of the AI launching them increases - but only if it makes sense to invade at that time - this is not a railroaded system, it is not meant to make those invasions happen in the same order on the right date.
* The leader of the operation brings in all the faction members for a coordinated invasion if possible, focusing their fleet and air operations around the invasion zones, and prioritizing the landing fronts for force concentration and certain objectives around the landing area for AIFC targets.
* It will focus on one of these operations at any time, with a faction coordinated (to a degree) planning, execution and rapid phases (i.e. less likely for the AIs to launch invasions one by one and fail each) and tracking the progress of the invasion for deciding whether to continue or abandon it. The AI's will synchronize the front-wide operational offensives in the landing zone as well.
* Added some special strategies for ENG not to overload likely invasion staging areas, and to construct infrastructure and naval bases in them.

Axis invasions

* Similarly, operation Sealion and the invasion of America (and Japan's invasion of Australia) are handled under this system, however unlikely they are.

Pacific War

* The Pacific campaign for Japan and USA is guided by another specialized AI system. It aims to combine island hopping, fleet operations and air forces into one cohesive strategy where the AI sweeps through the Pacific in a more deliberate way, focusing forces and securing objectives within certain sectors of the Pacific before moving further, without being railroaded to some specific order of islands to go through for example.

Note

* These systems are ALL experimental beyond making them functional. You are the tester.
* Also, this does not mean the AI has gained some incredible ability to micro its way around difficult situations (which invasions etc. usually are). They guide the AIs to cooperate with each other, or coordinate and focus their own forces into doing certain things.

Germany

* Reworked advisor pick order, national focus order, some decision logic, priority research, production, new mechanics etc.
* As usual, it will go for a strong air force, but it should have a lot more armor divisions than before.
* General diplomacy related focuses and such are meant to be fairly historically accurate with timing if nothing terrible happens, with a couple of exceptions like the invasion of Norway and Yugoslavia. * Reminder that this mod is not a 100% historical imitation and there is no guarantee that in a random game the AI doesn't get into some shenanigans (even without you pulling off meme strategies) that cause severe delays to historical operations and I won't railroad the AI into doing so.
* MEFOs: optimizations done here but will probably need further tweaks
* Germany does collaboration ops in Norway to make capitulation easier
* Will try to not spend its entire air force on Battle of Britain before Barbarossa
* Sync sealion/atlantic focuses with naval invasion ops
* Has prio strategies for developing rockets, researching the relevant techs, building some rocket sites

Soviet Union

* Focuses on making a very large army at first. It will only start up armor production if it can afford to. If you want it to field a horde of tanks, tough luck! Give it buffs.
* Added a specific prio construction strategy for refineries during rubber shortage
* Fixed random issue with AI/USSR running out of manpower because it took total economic mobilization while thinking it could also get women in the workforce but randomly lost the war support for it while waiting for PP and got stuck in a bad situation... Now it reserves enough PP to grab both at the same time and won't take total mob. unless both are available. If it can't get women in the workforce due to war support while also wanting total mob., it will reserve and use PP for propaganda decisions and then simultaneously take both total mob. + women in the workforce... no way this could go wrong
* Added some emergency production strategies for it to shift factories to infantry or air equipment from armor production when having shortages in those areas
* Tweaks to defense related decisions
* Re-enabled and expanded detection radius of all defense-in-depth strategies, for several more key cities for the AI to guard with troops when enemy is near, and gave Leningrad special consideration for the AI

Japan/China

* China doesn't rush to take away Indochina from AI Japan
* Significantly refined the stalemate setting AI for China/Japan and made it the default setting
* Fixed some early game factory assignments
* A lot of front strategies added/reworked for a more coherent overall strategy in China
* Made sure that it will research the SHBB project, wait until the focus is complete before queuing SHBB, and prioritizes finishing the focus inserted SHBBs
* China's United Front forms properly
* Optimized Manchuria's focuses/decisions/banditry (though vassalize MEN path is still blocked for AI)

Japan/USA

* Increased island garrisons

Special projects and facilities

* Radar, super heavy battleship, rocket techs have some special prioritization in this mod to make sure the AI saves points for those if it wants them
* Added priority construction strategies to guarantee the bigger countries get relevant facilities

Templates

* Corrected a lot of mass assault infantry template widths and added cheaper versions for poorer countries
* Armor template scripts have been reworked again to allow the mod systems to work. The AI won't use 3 types of brigades in a template anymore, I don't know how to make it do that successfully anymore so don't ask. On the other hand, I am letting it upgrade to mech/modern types again, since there are ways to make it a little less painful for the AI now. Also, bringing back light flame thrower support with fuel drums since some issues with factory assignment can be fixed now.
* Made sure that all but the cheapest main infantry templates the AI uses have some sort of anti-air
* Marine templates the AI uses are always 35w if possible, and use pioneers instead of engineers if available

Naval invasions

* Found a few tricks to have more control over the AI's naval invasions, so now you may see massive coordinated invasions again where the entire Allied faction joins in on targetted invasions for example as it did years ago in this mod.
* Naval invasion coordination action: may work better now but its up to you to find out.

Misc

* Minors more pre-emptively raise conscription against bigger enemies
* Improved how war trend is calculated
* Improved Spanish civil war and North African AI behavior scripting that nudges it towards historical results
* Increased armor division production/tweaked factory assignments
* Fixed an issue with AI selecting the wrong doctrines instead of those set by EAI (using the allow tech triggers to force the AI because it is otherwise uncontrollable)
* Infantry division related techs are generally prioritized more
* Aggressively prio constructs infrastructure in states that have a decent amount of a specific resource the AI needs.
* Added a number of checks and warnings that will surely stop anyone using the mod in unintended ways
* Disabled EAI managed army/air/navy XP as there should be no issue forcing the AI's xp use by other means
* Adjusted AI's MIO trait selection to prefer everything that makes production easier, attrition less
* AI assigns more dockyards relative the amount it has to floating harbors to build a decent stockpile (so a lot for USA)
* Researches improved transports and attempts to build them a little more, though I don't know how useful they are even now.
* Made sure the AI takes the spec ops doctrine branch it needs