Noticed a play the AI made a few times. Now playing 18.104.22.168.3266.
It seems pretty agressive with its tanks attacking "trading" territories.
Included a savegame where it attacked two territories with value 2 defended by 1 inf.
In both attacks it used 2 tanks. There were also 1 inf and some fighters available and not used that turn.
Both territories are vulnerable to couterattack so essentially sacrificing the tanks.
Whether the sacrifice is sound tactics in this scenario can be debated, i'd say it's not.
Think the cost outweighs the benefit. Not sure how the AI makes this consideration.
Have seen AI being generous with expensive tanks few times before.
The AI has a big issue, that practically bugs off its gameplay, when it is confronted by more than 1 territory with no income to attack, to go through.
I know that the AI in the past had such kind of an issue for whatever territories worth nothing, and it still appears having something the like in case of multiples.
As you can see, in the savegame below, the AI has packed a total force of 5 spearman, 1 archer, 1 swordman and 1 cavalry in Thenae on round 3, and it appears never taking the decision to push through the east, taking Sabrata then Oea; in round 9 those forces are still sitting there and, I guess, will stay there forever, unless something else impacts thereabout.
I think this is almost a bug, and practically means that whatever maps having two serial value 0 Neutral territories to be taken to proceed is basically wasted for AI gameplay (think about situations like 270BC Paraetonium, setting apart Egypt and Numidia, but with two or more of them, instead of only one).
In this mod, Numidia is indeed supposed to move most of its forces in the western half to the east pushing through western Syrtis, and it appears unable to do this, while somewhat understanding it should do that, since it packs them in Thenae.
Side note, I was Carthage only (all rest to Hard AI) and the Hard AI gameplay was very bad, allowing me to take Roma on round 5, 7 and 9, plus Sparta on round 9, as well, but this is mostly related to its apparent inability to thinking ahead, so that I just need to assure that the next round I will be able to mount a pressure that it can't face with its latest turn gameplay only. For Roma here, a simple solution would be making the AI able to purchase and place Forts, as it can put 4 per turn on Roma (while it is complex to make the AI choose immobile units, I guess it should be fairly simple to just patch it to buy them as much as that is the only way to defend a capital, which can almost never be a wrong decision).
Also, there is a cavalry of Numidia that appears lost. You can see that cavalry that took Siga on round 2 is still there on round 9, going nowhere. It should either attack Banasa (not a good attack, normally, but better than being useless) or move eastward, to push through Sabrata and Oea, as well (I would move eastward).
Also, the bireme in 29 looks lost, as well; it never moves for the whole game at all! It should be fairly obvious that it should move eastward and do some disturbance against the RomanRepublic ships. Also, it can be used to ship Numidia units (usually not more than 1) to garrison Caralis, and eventually menace some landing against RomanRepublic.
Yeah, not really important; that mod I've yet to try to add to the repository might be the only 1 game around in which you have two value 0 neutrals to go through. Also, I recall that I've seen the AI taking value 0 neutrals for no reasons (they were not needed for movement or anything). So, I'm not sure under what conditions the AI will want to take a value 0 neutral. Anyways, taking useless vaule 0 neutrals is a marginal issue.
Just wanted to post up a notice (again), in case this matter has been forgotten, that, when you removed Moore AI, this had the effect that a few maps, that were and are meant to be played with AI, are now not anymore balanced, and, since it is not explained anywhere that they were balanced for an AI that doesn't exist anymore, people of course think that they are meant to be played with the current Hard AI, as it seems the only logical thing to assume.
This goes back to my suggestion of having an "Outdated / Broken Maps" category, in which to move such cases, and maybe tell what is the problem in the download description, waiting for anyone wanting to take them over and attempting rebalance, but I know my suggestion of having a category for Broken maps has been refused, already.
Whatever the lead developers think is best as a solution (the solution I gave was making a "Broken" category in DL list, but was not accepted), I guess it is your call; I don't play any of those maps, but I think they should not be removed from DL list either. Main point is not making people wasting time by making them think that those maps are meant to be currently played with AI, while they are just meant to be played with an AI that doesn't exist (pretty much everyone will totally waste time trying to play them with Hard AI, because how is he supposed to know???).
Captain_butter: have anyone tried to play eastern front against hard ai
Captain_butter: damn near impossible
Captain_butter: never succeded
Cernel: you mean you won?
Rogue_Two: challenge accepted haha
Captain_butter: rus have like 500 ipc
Cernel: nearly or absolutely impossible?
Captain_butter: seems impossible
Cernel: You just slapped Maxwell-Lord
Cernel: probably it is
General_Zod: 1 ai enemy should be easy
Captain_butter: unless one takes all factorys from rus
Cernel: that map was meant to be played with the old Moore AI
Cernel: that has been remove
Captain_butter: but still impossible
Captain_butter: cause i started to feel really dumb since a lost every time against ai lol
Captain_butter: so that map is officially corrupt?
Cernel: yeah, it's kind of bad actually, because people can't know that one and several other maps were balanced for an AI that doesn't exist anymore
Cernel: so I guess people will think that they are balanced for the current AI and get frustrated
General_Zod: those maps should be but in a new category, broken
General_Zod: not experimental