AI observation and advice for improvement

Moderator: Hubert Cater

Post Reply
Xsillione
Posts: 212
Joined: Fri Jan 18, 2019 9:36 am

AI observation and advice for improvement

Post by Xsillione »

Played a Japan only game (still playing, '43 sept), but already has a few strangeness about the AI:

1. The AI will always use the airforce first, while this saves the land units, as softens up the targets, also waste some of the damage to entrenched units, and generally wastes airstrikes. Also makes it impossible to strike the damaged and near destroyed units into nonexistence for greater economic damage and allow deeper land access.

2. The AI will waste airstrikes on surrounded units and irrelevant naval units, instead of actual important targets.

3. The AI regularly moves before attacking, even if they attack the unit that is next to its starting position, losing the unmoved attack bonus and some movement point.

4. The AI will always use its frontline units first, so when attacking the Benelux, France, or even the SU, they move the foremost units, and than cannot move the second or third row into positions, simply because they cannot move far enough, or all the places are locked by the already moved ones. If used the backrow first, they could later move with the original frontline further and do more damage.

5. The AI almost never do opportunity strikes against vulnerable naval units that are important. During the Benelux phase of the early war, half of the home Fleet was near the coast, since the German fleet also showed itself. Yet, only the soon to be surrendered fleets got destroyed, the permanent UK ships got only one strike in three turn, while attacking them would slow down the french attack, sinking 2-4 UK capital ship would probably worth it. Later in north africa, the UK placed a carrier into a port, that could be striked with an armor and an infantry, yet, all three attack went against the unit in the city, instead of the port with the CV, not even kicking out the unit.

6. The AI use way too much operations. After Poland, they operated third of the units, yet only attack in normal time. After the fall of France, they operated almost half of the units near Yugoslavia and SU, despite the fact, that it had 4-6 turn to walk for Yugoslavia,and 6-8 turn for the SU, no need for operation, unless they attack early. Also from Yugoslavia and Greece, the AI operated to the SU front, saving less than one turn in most case at huge cost.

7. The AI regularly drop paratroopers behind line without any chance for supply, just to lose them instantly the next turn. They not even dropped a city that happened to be occupied, just into the middle of nowhere behind line.

8. German AI Sent two Armor and three infantry to Turkey, with zero HQ.

9. Axis AI garrisoned 23 cities and partisan zones with corps and armies instead of garrisons, most of them fully upgraded German units, instead of Italian or minors. Yet no invading force nearby or even threatening anywhere. Also 4 airunits are helping with this garrisoning.

10. Almost all German units have mobility upgrade, since four army with mobility cost as much as five without it, it is quite a waste, when they don1t really need it. Even worse, the garrisoning corps and armies also have this, for standing and not moving at all. Also, minor units should be upgraded faster, while a bit ahistorical, it is the only logical action with the current rules. For a tiny bit of investment, the minor units could be much better.
User avatar
Hubert Cater
Posts: 5103
Joined: Mon Jul 22, 2013 11:42 am
Contact:

RE: AI observation and advice for improvement

Post by Hubert Cater »

Thanks and for 3) it doesn't actually cost the AI anything in terms of unmoved attack bonus. What the AI does here is it is usually trying to get closer to its goal position before it attacks and sometimes, even if it is already next to its goal position it might move a bit first before it attacks to open up that hex to other units to also attack. The thing is, in this latter case, it will only do this if the expected combat results will be the same if it were to move or not move. Sometimes this is the case, where attacking from one hex with the unmoved bonus will be let's say 1:2 and moving to an adjacent hex and attacking will also be 1:2. At least it should be as this is what I have the AI checking for.
VSente
Posts: 13
Joined: Fri Aug 11, 2017 7:40 pm

RE: AI observation and advice for improvement

Post by VSente »

Point 10 also seems very important to me.

In a similar game with AI-controlled GER all of its land units had improved mobility.

(I'm sorry I have a screenshot but I can't upload it [&:])
Mercutio
Posts: 255
Joined: Mon Dec 25, 2006 11:49 pm

RE: AI observation and advice for improvement

Post by Mercutio »

With the constrained map in Europe vs WiE, the mobility upgrade seems a bit wasteful IMO as VSente points out. Maybe a few spearhead units, but certainly not worth throwing the MPPs around over.
Captjohn757
Posts: 102
Joined: Sat Jan 07, 2017 2:17 pm

RE: AI observation and advice for improvement

Post by Captjohn757 »

Well, if you find the mobility tech wasteful, it's certainly your option to pursue it or not. I concur that the confines of Europe have a tendency to mitigate its value from a limited perspective, but not necessarily so in southeast Asia where terrain, jungle, mud, rivers, rain and zones of control considerably inhibit movement fluidity. The extra action points are a boon to operations in that theater.
Mercutio
Posts: 255
Joined: Mon Dec 25, 2006 11:49 pm

RE: AI observation and advice for improvement

Post by Mercutio »

The problem with SE Asia is you have to research it. Japan has enough other things it needs to research and buy. China as well. India and the UK are the only ones that could probably make that work IMO.
Post Reply

Return to “Strategic Command WWII: World at War”