Athena definitly gets the reckless perk in my books

[pepe4158]

Say Pocus, the new Athena is a bit better, but anyway you can turn her aggression down a bit...shes border line suicidal sometimes. (Barksdales noticed it too in our e-mails)
Her worst move seems to be to way out run her lines of supply, and want territory way too far forward she cant support...any way to turn that down a tad at some further time?
I put her aggression down all the way...doesnt seem a difference?
I wasnt sure where to put this wish and observation...hope I got the right spot?

[W.Barksdale]

I think the problem is that Athena "sees" where your forces are not rather then where they are.

[Pocus]

Good point and right place

I have not tweaked her recklessness, but it can be a ripple effect of another change.
I would need some help from you, it saves me time better used for the actual tweaks. A precise example of recklessness, or a stupid winter move from her for example. Analyzing the stack order, I can know why it was orderered such move.

[GShock]

Send/attach savegame / pics (Use fraps if you need http://www.fraps.com) pepe :-)

[Clovis]

I've tried with normal aggressivity and small Fog of War advantage with the 1.10: CSA AI was targetting Boston with JE Johnston 's Shenandoah force during the summer of 1861.

Another curious move was BEAUREGARD going to Fort Monroe with all his army in the same turn, letting path to Richmond totally open to McDowell.

AI doens't seem to appreciate both supply situation in the targeted region and attrtion and cohesion losses in order to reach the targeted region.

AI is too certainly praising too much offensive against weak points against defensive necessities. In some way, I've now the impression AI tailored to take offensive somewhere whatever the situation rather than keep a defensive stance.

[Carnium]

Another curious move was BEAUREGARD going to Fort Monroe with all his army in the same turn, letting path to Richmond totally open to McDowell.

I can confirm this too as it happened with game patch 1.09E too.
CSA aggressiveness was set to normal and all other rules were left to default.
Is there a way to improve the defense of Richmond with a special garrison which would spawn when the US forces are near or would it be too ahistorical ?

[Pocus]

I've tried with normal aggressivity and small Fog of War advantage with the 1.10: CSA AI was targetting Boston with JE Johnston 's Shenandoah force during the summer of 1861.

Another curious move was BEAUREGARD going to Fort Monroe with all his army in the same turn, letting path to Richmond totally open to McDowell.

AI doens't seem to appreciate both supply situation in the targeted region and attrtion and cohesion losses in order to reach the targeted region.

AI is too certainly praising too much offensive against weak points against defensive necessities. In some way, I've now the impression AI tailored to take offensive somewhere whatever the situation rather than keep the defense.

Some good pointers. Now it has to be translated into something intelligible by the AI.

[Clovis]

Some good pointers. Now it has to be translated into something intelligible by the AI.

and that's is really difficult.... I know.

For now, I commend to play with low aggressiveness and no Fog of war advantage setting. For Federal AI, a initiative bonus seems to work better...

[arsan]

and no Fog of war setting

You mean no fog of war advantage (standard settings) or full advantage (no fog of war at all) to the AI?

[Turbo823]

Here are some simple ideas that I think can help Athena's play:

Use VP cities to help Athena with sound routes of supply.
I've been experimenting around with VP and I think using the victory points is a good way to help Athena play more soundly. Think of it as laying bread crumbs for Athena to follow.

Stronger garrisons for strategic and objective cities:
Losing weakly defended objectives or strategic regions really seems to disrupt Athena's supply lines and cause her to do bizarre things such as moving in winter out of supply. I think that increasing the size and strength of locked garrisons in key cities just for Athena helps her protect her key objective cities. Do the same for strategic cities. If the city is re-captured, give Athena the garrison again and again.

Securing supply lines: Garrison all forts and cities!
Athena often leaves settlements undefended and these that revert back to enemy control. I think that giving Athena a locked garrison militia (even as a gift) for every settlement/fort she controls (even those that are in Enemy territory) would help her play immensively. This has a very positive impact on Athena's supply lines when she is deep in enemy territory.

[W.Barksdale]

I think the problem is that Athena 'sees' where your forces are not instead of where they are.

[Clovis]

You mean no fog of war advantage (standard settings) or full advantage (no fog of war at all) to the AI?

no fog advantage. with, she sees your weak points in your rear and tries to capture them. Without, she's more shy about locations without any intel data.

And even the smallest fog advantage reveals a large part of the map.

[arsan]

no fog advantage. with, she sees your weak points in your rear and tries to capture them. Without, she's more shy about locations without any intel data.

And even the smallest fog advantage reveals a large part of the map.

Thanks. Yes this is how understood Athena worked.
Thats why your "no fog of war for the AI" sounded so strange to me
Cheers

[mikee64]

OK, guess I will use this thread in hopes the "help improve Athena" efforts will continue but not clutter up the official patch thread.

So, here is the situation, same game I posted about in the patch thread:
In the East, Banks ~130,000 man army reoccupies Manassas in an advance from Washington during early February. Jackson had marched in, burned the depot, and marched back out while US forces rested in Washington the previous turns. The main rebel army (HQ + 3 corps) occupies Fredericksburg with a strong supporting corps in Charlottesville. So far so good.

Late February: Banks attacks Fredericksburg, crossing the river, and suffering a 35% command penalty. Despite a 2:1 manpower advantage, the combat value favors the south something like 2000:1600. Banks is defeated suffering ~20,000 casualties to ~5,000 for me. Not too bad, actually not too far off from the results of the "real" battle of Fredericksburg in abstract game turns.

Early March: Here is where I start to have a concern. Instead of withdrawing to Manassas or elsewhere, Banks stays in Fredericksburg, still with 35% penalty, risks being caught outside in bad weather, and attacks again with literally no chance of success. Results:



Same turn, Halleck sits out in the open in SW Missouri with a decent sized force. Although the weather is clear this turn, he was also there last turn when it was snow, and obviously risked a snow turn again:


I guess we'll see his intentions next turn.

Pocus: save files have been emailed.

Edit: here is the recon on Banks' force:

[mikee64]

Athena got her act together somewhat over the next few turns. She got Banks' army down to only a 10% penalty by forming several corps and adding some signal units. Out west things are quiet except for Grant who has taken a 12,00 man force deep inside my territory and is besieging Chattanooga after a failed assault against a 6,000 man force. I see no other real threats at this time and am mostly ignoring things waiting for Athena to make some moves.

With her mind obviously set on Fredericksburg, she attacks here again with a global combat value of around 2,800. Surprisingly, Beauregard (Army HQ 400 GCV), Jackson (loaded corps with 1200 GCV), and Bonham (corps with 600 GCV) all decided to retreat west to Culpeper w/o firing a shot. They were at entrenchment levels of 2/3 and all set to standard defense (blue/orange), outside the structure. This left Johnston, who was serving as my "hospital corps", besieged inside the city with a GCV of ~800 and an entrenchment level of 5. I thought the retreat was interesting given the "big battle" thread over on the other topic (more on that follows).

The next turn I order Beauregard, Jackson and Bonham back into the region to attack Banks outside Fredericksburg. This will be a coordinated attack but across a river; a risk I'm willing to take against the AI. This is a move I would not even consider in a pbem, but I find it makes play more fun against Athena to make a few concessions. Lee is also ordered to attack with a newly formed corps moving north from Richmond, leaving the capital weakly covered. He will arrive after the initial attacks from the main army.

What follows is a BIG, casualty heavy battle. (Since this is day 1 it looks like my forces marched to the sound of the guns). I lose 2 NM from this one:



Day 2. I gain 7 NM from this one:



Final day, most of my punch in this one comes from the arrival of Lee:



Too bloody?

[berto]

Too bloody?

In your example, both sides suffered more than 50% casualties over the three days of battle. Given that the bloodiest battles of the Civil War were far less costly (30% Murfreesboro, 27% Gettysburg & Chickamauga, 20% or so for the other "bloody" battles), yes, I'd say about your battle: too bloody.

[Banks6060]

Yeah I've been playing the '62 scenario as the rebs. Interesting coincidence with mikee's game....Banks takes the head of the main Union Army and runs it into battle without first forming a decent Corps structure.

Somehow I think this game may be too complex for even the AI to figure out

[Fern]

Too bloody?

Union:
80.000 KIA. WIA or MIA out of 130.000 (61%)

Confederates:
42.000 KIA/WIA/MIA out of 67.000 men (62%)

Yes, I think it is way TOO blody

[Gray_Lensman]

deleted

[berto]

Clovis has had some luck adjusting the cohesion losses from battles that help to lower the casualties and increase the chance of routing. Hopefully, this can eventually be applied to the vanilla game also.

This factor alone impels me to try his mod (for the first time).

I, too, would vote for these adjustments going "vanilla".

[mikee64]

Clovis definitely has some great ideas and has done a great job on the mod. The only reason I am avoiding it is because I want to have clean install save files to send to Pocus. In principle I absolutely agree with his mod approach.

Athena got a bit restless with a 15,000 man corps under Butler while the rest of the Union army was recovering in Washington. Butler tried to cross from Culpeper to Albermale, in between the two wings of my army. Unfortunately for Butler, Lee was moving into the region in attack posture. The results were not good, as the river prevented a retreat:



Question is, why was he even moving there in defensive posture?

Grant withdrew after the failed Chattanooga raid; there doesn't seem to be much happening in the west at all with the AI while tons of resources are obviously being used up in the stalemate in VA. McClellan does have a decent sized force wandering in MO but I don't see any tangible objectives he can take there.

Another thing Athena is not doing a good job of garrisoning her forward depots at all. I'm not doing any deep cavalry raids, but I am probing to locations that can be reached in one turn. Examples of areas with unguarded depots now occupied by raiders are Ashland, OH & Salem, IL.

She also loves to build supply wagons (and print money; I assume the 2 are related given the cost) which is fine. However, she has a bad habit of letting the wagons travel unescorted through risky areas, where they are easy targets for capture by cavalry and/or Indians.

[Clovis]

Another thing is Athena is not doing a good job of garrisoning her forward depots at all. I'm not doing any deep cavalry raids, but I am probing to locations that can be reached in one turn. Examples of areas with unguarded depots now occupied by raiders are Ashland, OH & Salem, IL.

I will implement the autogarrison feature for AI this week in my mod.

[Brochgale]

On subject of supply wagons - I just grabbed 5 in one turn from Feds strolling aroung Fredricksburg. And JEB has not even appeared yet. Or maybe he is ghosting in the AI?

[mikee64]

The disregard for security of supply wagons is a major fault of the AI right now. It had happened many times before I reported capture in my last post. I've got so much captured supply right now I don't know what to do with it.... and I am playing a very conservative game.