Hi!
My experiences with the Leader mod are very similar to de ones posted by Franciscus as I stated on the other thread about this
http://www.ageod-forum.com/showthread.php?p=70770#post70770
I don’t believe it is because of the relocation of the one star leaders on this mod, but because of the historical changes included on the high level leaders (** and ***).
I think Athena needs help forming armies, whereas it is getting the armies already formed (like on the main eastern armies on USA and CSA on the vanilla game) or at least receiving HQ and *** leaders at roughly the same time and on the same place.
She has no trouble making divisions anywhere, so I think we could use the relocation events for the divisional leaders without problems.
I hope Pocus finds time to continue working on Athena but meanwhile and even in addition to AI improvements I think that creating (or modding) an AI friendly setup will be a very good addition to the game, as most people plays against the AI.
With this I mean a slightly different setup for each side, tailored for use when the AI played that side.
In this setups I will include some 3 stars leaders already attached to a HQ (for example AS Johnston could enter with his army, and maybe Buell too. I think Clovis already does this with Johnson on his mod). Also, some arrive dates and locations could be changed to help the AI, following historical guiding lines (for example, McClellan could enter later but already on Washington DC or Bragg around Nashville and not down on Florida)
The changes would be aimed at giving the Ai some basic and historic Army and leaders structures on which she could develop further strategies.
Its something the game already does for both sides on the east at the beginning of the campaign, but not on the west or transmississippi theatres.
Other improvement for “AI only use” could be providing a new system for the neutrality of KY. As it is now, Athena will always lose the state by attacking first some stupid militia on Paducah or Lexington, except if the human player does it first on purpose.
For playing against the AI it will be better to use a simple % probability event like the one used with Missouri loyalty (St Louis Massacre). This way, at least one will get a little variation from game to game.
For PBEM, the current system (or one better…

) could be used.
Has this been thought already?
What do people think about it?
Regards!