Observation and opinion....

[gbs]

Now that I have almost finished the 61 July campaign as the CSA (turn 99), i thought I would make a few comments on things I observed. I have the following settings:

Version 1.03
Difficulty level - Normal
AI Detect Bonus - No bonus (I done want AI cheating)
AI Agressiveness - High
AI More time - No

After taking Washington in late 1864 my NM reached a high of 152 while USA hit a low of 58. My NM has backed off somewhat to 142 as i have made some command changes that the public did not like. I have 4193 victory pts. vs 3439 for USA.
I control every major city in Kentucky. After I took Louiosville and Lexington the union Army placed what looked like large forces all along the the Ohio river. I dug in waiting for the attack and after two years none has come. The only union activity has been raiders sent into Tennessee and Mississippi causing minor problems. I have never seen GRANT!
I was basicly run out of Missouri, all the way to Ft Smith. The union army took Fayetville but stopped there. They have been sitting for over a year now with no movement and I have been unable to dislodge them with two divisions.
I am currently forming a new Army of Mississippi in Memphis, commanded by Bragg with three corps. For the rest of the game I hope to take St. Louis with this army along with elements of the Army of Tennessee in Kentucky since the AI doesn't seem to want to fight in that state.
In the east there has been no offensive action by USA since the first year of the war. That was mostly in the Valley and the pininsula from Norfolk. After I retook Norfolk and took and held Harpers Ferry I have received no more attacks there. The USA did probe a bit at Manassis but after beeing whipped they moved back to Fairfax where they settled in until eventually being evicted by Lee. I now have a huge Army of NV under Lee (200,000+) with five Corps in Washington. There seems to be an enormous Union presence in Annapolis and Baltimore. I an currently planing an attack on Annapolis.
The AI seems to be playing defense. In reallity, if this were so The USA would have sewed for peace long ago rather just wait for an attack by CSA. If memmory serves the CSA had no designs on land above the Mason Dixon line, they just wanted out.
I have had great fun (dispite numerous crashes) and look forward to starting a new game. As you can guess, my biggest disapointment has the the lack of any kind of offensive action by the AI for the last 60 to 70 turns or so. I know this is just an AI and and have found over the years how difficult it is to teach an AI offensive warfare. I know nothing about programing. Pocus has stated many times how much he wants to spend time with AI improvement. I hope this is being done now and I hope there will be some future patches that address this as I want to start my new game against a more aggressive AI.
Anyway this is what I have found. I welcome any comment that anyone wants to make.

Greg

[Jacek]

Gbs, great to hear your story.

I play with following settings as CSA
Version 1.02
Difficulty - Normal
AI Detect Bonus - Low
AI Aggresiveness- Normal
AI More time - no
AI Behaviours - All

In Virginia AI is fairly good because of its numerical advantage. It probes in Norfolk region. It dispatches small forces in Mobile area. Less activity in the West. The way I see it is AI as the Union may suffer from lousy generalship (too cautious, big stack with combat penalties).

[gbs]

HMM. Interesting comments. I agree with lack of agressiveness early in the war. A more agressive union could have ended the war early indeed. But, what about Grant in mississippi and Sherman in Georgia later in the war. Like i said, I am in April 1865 and have yet to even see Grant and Sherman is somewhere in Maryland.

[Pocus]

I can just say that the AI won't do what you would not do, ie she will not try to attack unless she can win the fight. Why the Union would send units to be butchered against Lee and 5 optimized corps, knowing that the chances to win are nearing zero?

I'm sure that, as a player, I would be able to bother you, like making operations in unexpected areas, or landing on your back, but for now the AI is not able to do that if you 'neutralize' her by having enough troops in front of the main cities to be guarded.

[gbs]

Pocus, please don't take this as an attack as it is not meant to be that way. As I said i have had "great fun". Thank you for that. With all due respect, an example. I took Kentucky and was holding it with a much smaller army of tennessee. It looked like USA had huge forces in the area and most likely would have defeated me if they had wanted to. Maybe I should play as the uSA once to get a better feel for what they have available. I just thought that the USA outnumbered the CSA by a good bit.

[Pocus]

sorry for my tone too, I did not meant to be aggressive either It is a bit frustrating as of now to not be able to pour as much hours as I would like into the AI... but I'm confident that things can be improved much more in the following weeks.

[Adlertag]

I'm sure that, as a player, I would be able to bother you, like making operations in unexpected areas, or landing on your back, but for now the AI is not able to do that if you 'neutralize' her by having enough troops in front of the main cities to be guarded.

It is maybe a problem impossible to solve in any game with such a high number of turns like ACW's grand campaign. AI doesn't like big number of turns, task for it is growing too hard, no matter programmer efficiency (because I know Pocus is doing the best possible job).
So a campaign of 100+ turns is maybe a too big challenge for AI ( War in the Pacific, Matrix Games, is another example).

So the question could be : How much turns do we have (rely on)for an optimal AI efficiency before it will "sleep", irremediably countered by player's actions ?
The more the better, but how many ??

[anarchyintheuk]

Although mostly based upon BoA if you want the ai to play aggresively/launch offensives you have to give it a substantial detect bonus. Haven't really found a comfortable setting for ACW but I usually played BoA w/ the following:

Difficulty - High
AI Detect Bonus - Normal (whichever the third one is)
AI Aggresiveness- Normal
AI More time - yes
AI Behaviours - All

Never really noticed a lack of aggresiveness. Going to try this setting on my next game of ACW.

[gbs]

Thanks. I'll try my next game with those settings as well.

[Carrington]

It is maybe a problem impossible to solve in any game with such a high number of turns like ACW's grand campaign. AI doesn't like big number of turns, task for it is growing too hard, no matter programmer efficiency (because I know Pocus is doing the best possible job).
So a campaign of 100+ turns is maybe a too big challenge for AI ( War in the Pacific, Matrix Games, is another example).

So the question could be : How much turns do we have (rely on)for an optimal AI efficiency before it will "sleep", irremediably countered by player's actions ?
The more the better, but how many ??

I've been playing computer games for 15+ years, long enough to realize that A/I will be truly artificial. I know it's not a particularly constructive point, but I keep coming back to the realization that these games are best played against another person.

Playing against the A/I for me is a lot like putting together a puzzle. Granted, some people like the big, 1,000 piece puzzles, but I find it unrewarding.

Indeed, more often than not with these big games I end up "hotseating" against myself, just because it seems the best way of figuring out learning what's going on under the hood.