Chaotic force structures of the Union.

[veji1]

Hi, I just wanted to rant a bit regarding a very frustrating aspect in the game. I bought CW2 a month ago after finallly updating my old laptop and some aspects of it have been very frustrating as an old AACW player. Particularly to see that the AI still has the same major issues regarding force organisation, namely a massive lack of division forming.

I tend to play the CSA and it has been maddening to see the Union have big stacks of 3000 or more CP but with just One or 2 divisions and tons of brigades. It makes their forces so much weaker. This is compounded by the fact that when one doesn't own the DLC, one cannot start the campaign in spring 62, which was in AACW the best way to get a somewhat fun experience and balanced game as a CSA player against the AI, because the force structure of the AI was already set and much better.

This has really frustrated me because it ruins the game : once divisions and corps are there, having an AI that doesn't know how to organize them sucks. To be honest some form of cheat "code" that allows the AI to generate default officers to create divisions everywhere (for no cost) would be much much better, it would just have a check at the beginning of each turn to replace a generic leader with a "named one" in any possible case.

Second problem, the AI just doesn't know how to attack adjust its routine around Washington, (btw I play with lieutenant difficulty, moderate detection and activation bonus and normal aggresivity). In summer 62 i went for a classic text book left swing with Jackson and Lee through Mongtomery, MD to get Baltimore and Annapolis, leaving Longstreet with 2000cps in Manassas and Holmes with 400 in Fredericksburg. Lee and Jackson each have 2000/2200. I can see this massive stack of 8 000 cps in Washington and 4000 in Alexandria, and they don't move. No big attack on Annapolis where Lee is stationed while Jackson occupies Baltimore. Worse than that, whereas the AI just waits and waits with its big masses, a steady stream of brigades arrives piecemeal at Annapolis (disembarking from boats), probably trying to concentrate in Washington and keeps getting annihilated... after 3 or 4 turns like this I just abandonned the whole area and retreated all the way to Culperer and Strausberg to try and get the AI moving. This is very frustrating because the AI had decent odds : a 6000/7000 CPs attack on Annapolis the turn after I had arrived would have been a touchy affair : with Lee + Jackson MTSGing I would have had around 4500 CPs not very well entrenched. Even if the Union lost, it would have hurt me making me consider retreating back to the south. By the way no Union force was remotely getting organised north of Baltimore, it felt like I could have walked all the way to Phila and NY....

I must be honest, I am really frustrated with this game : many interesting ideas (Regional decisions notably), but not much progress in the wargame itself. I know the AI will never beat me under normal circumstances, but it should at least be able to form a coherent army even with "cheat generals". To see stacks of 3* generals roaming around in the west or Transmississipi with 1 division and 4/5 or sometimes more brigades on the side just sucks. I had only one somewhate realistic battle with Grant/Halleck leading a 3 division force between Munfordville and Bowling green, otherwise I keep chasing shadows or wiping the floor with badly constructed stacks. And I have been careful not to "shock and awe" the AI, roleplaying some what, for example not attacking Louisville and Lexington, leaving my army of the tennessee in Bowling Green. With its detection bonus any semi apt AI should have taken Munfordville in force and camped itself accross from Bowling Green. Only in Missouri has the AI been somewhat active,sure sometime incoherently, but at least I can rationalise and role play what is happening : I have lost and retaken Jefferson City and Rolla a few times, even Springfield once, I have taken Bloomington with partisans, etc.. The war is fluid and fun in this area, but got the west and the east are really stuck, I can't find a way to make the AI do something probably partly because of its very very bad force structure : It sees a stack with 3 divisions and no command penalty and won't attack it despite having the equivalent of 8 or 10 divisions because it just won't build them and then consolidate them into corps..

And to think that I would have to buy the DLC to get the workaround of the Spring 62 scenario where the Union's forces structue gives it the ability to move is maddening.

Really some cheat codes should be enforced just for the AI to help it structure its force :
- generic leaders to help form divisions every time there are more than 2 units of the same type (applies to infantry and cavalry, normal artie being "seen" as infantry ad horse arty as cavalry)
- annual or semestrial "C&C cleanup" where the AI is allowed to instantaneously move its leaders around to reorganise its forces (say in the middle of winter, make the troops concentrate in depots and make the AI organise its leadership coherenlty (at least by AI standards).

really that force structure issue has, as you can see, frustrated me to no end.

I hope something can be done.


/ Rant.

[Rod Smart]

turn up the difficulty
the forces in Alexandria and Washington are locked and cannot attack
if you want to take over the Union and reorganize them, then for one turn (probably in winter, since nothing interesting would be happening), take over the Union and reorganize them.

[Gray Fox]

You might play at Colonel difficulty or try PbeM.

[FightingBuckeye]

You might play at Colonel difficulty or try PbeM.

I wholeheartedly endorse the PBEM route. I just recently started PBEM and it's night and day how much more challenging and enjoyable playing against a human player is. No more holding myself back or limiting my options because Athena wasn't able to respond well to a given situation.

I shared many of your same frustration with Athena at times. It's a great game engine and is fun to play, but the AI is somewhat lacking. To be fair though, that's often the case in computer games. I have yet to play a computer game where the AI is competent enough to go head to head with a human player without some form of 'cheating' to give them a leg up. It kind of defeats the purpose of playing a historical game with tons of accuracy if the only way Athena can be a challenge is if their industry can actually compete directly with the Union industry. Give PBEM a shot and I bet you won't regret it.

[veji1]

turn up the difficulty
the forces in Alexandria and Washington are locked and cannot attack
if you want to take over the Union and reorganize them, then for one turn (probably in winter, since nothing interesting would be happening), take over the Union and reorganize them.

The problem is that the AI just stacks all the troops together in Washington and Alexandria therefore locking itself up.

But I suppose you are right, the best thing to do would be for me to load up the game as the Union in the winters and clean up the logistical mess.

Re PBEM, I know guys but nowadays the way I play computer games is on the go, when i am travelling for work, or in a sudden intensive sitting when I have a free evening and play 3/4 hours straight, than none at all for a few days or weeks. It really doesn't fit well with PBEM.

The reason I am mad is that I don't expect the AI to be smart of course, I don't expect it go on clever offensives, etc.. but at least it should be able to structure its force decently enough. Not optimally mind you, but just decently : build divisions and build some corps, even if they are just used together with the army in one big stack....

I can live with a lacking Athena : I tend to role play quite a lot so as not to abuse the AI. But when the AI just gets bogged down by it inability to organise its forces, it drives me mad..

[dinsdale]

You may want to check out this thread http://www.ageod-forum.com/showthread.php?39384-Is-it-me-or-in-1-05-is-the-CSA-smaller-more-passive and try 1.06 beta.

Pocus's AI's have been good, going all the way back to Birth Of America, I think if there are issues this major then the answer is not to turn up the difficulty or play PBEM, but to see what the problem is. While not human, the AI has always been able to play according to the rules, manage its armies and provide a good defense.

[Taillebois]

Try Command Ops. Good AI. No cheating. On bargain offer from Lock N Load for a short while. Personally I always seem to need a stupider AI opponent.

I wholeheartedly endorse the PBEM route. I just recently started PBEM and it's night and day how much more challenging and enjoyable playing against a human player is. No more holding myself back or limiting my options because Athena wasn't able to respond well to a given situation.

I shared many of your same frustration with Athena at times. It's a great game engine and is fun to play, but the AI is somewhat lacking. To be fair though, that's often the case in computer games. I have yet to play a computer game where the AI is competent enough to go head to head with a human player without some form of 'cheating' to give them a leg up. It kind of defeats the purpose of playing a historical game with tons of accuracy if the only way Athena can be a challenge is if their industry can actually compete directly with the Union industry. Give PBEM a shot and I bet you won't regret it.

[Gray Fox]

If I play the CSA, then I just take D.C. in 1861 and that's that. If I move the Union capital to NYC to prolong the game, then Athena's NM is below room temperature and her Divisions fight like they are with the YWCA.

If you play the Union against Colonel Athena as the CSA, then you get a pretty decent game. If you move the capital to Atlanta for Athena right before you take Richmond, then you can play on for a long game past 1863.

Some Divisions that Athena puts together are better than others, but a great deal of variety exists among human player Divisions too.

[RickInVA]

In a game I was CSA vs. Colonel Athena, in 1865 I checked the other side for leadership. Athena had 8 armies and these were the commanders:

Meade
Buell
McDowell
Halleck
Banks
Fremont
Ord
Rosecrans

Not being used:

Hancock
McPherson
Sherman
Reynolds
Logan
Slocum
Curtis

Grant seems to have gotten knocked on the head early in the game.

And this was not a situation where Banks and Fremont had become good generals. All of the "not used" 3-stars were better than all of the "used" except Meade. Additionally the Union had not been winning at really any time in the game, so it can't be assumed that no changes were made because they were winning.

I had hoped for better use of the better generals by Athena.

[Rod Smart]

I had hoped for better use of the better generals .....

Said Abraham Lincoln

[Stauffenberg]

turn up the difficulty
the forces in Alexandria and Washington are locked and cannot attack
if you want to take over the Union and reorganize them, then for one turn (probably in winter, since nothing interesting would be happening), take over the Union and reorganize them.

Yes and try maxing out AI advantages (actually a P.O.--Programmed Opponent)--in particular aggression to max, for a turn or two at critical junctures.

The Union PO should also be nailing down an effective blockade as well and I am not seeing this so far....

[ifailmore]

would sargeant ai plus high traffic and high delay movement, +1 activation rule be more balanced game? If i increase the dificulty it just increase the cohesion regain rate and movement which make it unbalance the ai is recovering faster or should i do colonel ai but no activation bonus with med delay and low traffic?

[Gray Fox]

I used the following settings for AI: Colonel, "Use all Behaviors" on, Normal activation bonus, the recommended low detection with normal aggressiveness and "Give AI more time" on.

[veji1]

I used the following settings for AI: Colonel, "Use all Behaviors" on, Normal activation bonus, the recommended low detection with normal aggressiveness and "Give AI more time" on.

Thing is this has no bearing on how the AI organises its troop. I personnally would favor a few AI "cheat scripts" activated at least once a year but ideally twice : once at the end of the late december turn and second at the end of the late february turn. Those cheat scripts would basically allow the AI to "teleport" for free leaders and troops to some second line bases and there make the AI "build up its forces" giving priority to the most able leaders.

Say for the Union AI the script would allow teleporting of "unconsolidated" troops (ie brigades, etc..) and leaders to some second line cities to allow "proper" build up of forces.

[Stauffenberg]

Needed for the US Programmed opponent ("AI"):

1) enhanced PO organization of bdes into divisions and corps

2) effective organization of serious invasion initiatives (x3: NC, SC & GA, and Gulf)

3) effective incremental tightening of the blockade

[veji1]

Needed for the US Programmed opponent ("AI"):

1) enhanced PO organization of bdes into divisions and corps

2) effective organization of serious invasion initiatives (x3: NC, SC & GA, and Gulf)

3) effective incremental tightening of the blockade

Man that would change the game... To me even if this required some cheating behind the scen (ie no teleporting of US units into CSA territory, but teleporting of units and fleets to the meet up port from which they set off the invasion, I am fully fine with that) it would make the game much more fun.

[minipol]

Pocus has said on several occasions we could help by creating "agents".
I'll quote Pocus from the message he posted on the 21st of November 2014:

If some of you want to help with a few AI scripts or 'agents' then that's quite feasible. Some betas knowing nothing about that did some good work in EAW, this is not technical at all. You just need a few hours to do improvements in the game, and there is no need for the developer here

[Mickey3D]

Pocus has said on several occasions we could help by creating "agents".

Have a look here for more information on AI agents.

[Gray Fox]

I certainly hope that more Database fields exist than the few that are listed. The LowHits and HighHits agents could form a unit with lots of hits, but it would just be a mob and lack any designed purpose. In the experiment I ran with Athena, half of the Divisions she had lacked even a sharpshooter and were mostly less than optimal. Can agents be created to use Priority for a unit (Division) to have not just a certain amount of hits, but to also have specific sub-units, i.e. a sharpshooter, line infantry or all artillery?

[Cardinal Ape]

Right from the start Athena knows that she is in trouble; she pleas to France for help by continually offering trade and territorial concessions. She sees Benjamin Huger in her database and thinks Lafayette will come to save her, but it is the wrong Huger of a war-game long since passed. (Lafayette stayed with Huger's grandfather upon his arrival to America.)

The AI should stop using those bad diplomatic options so often, possibly never.

I wouldn't mind if Athena received a blanket combat bonus at higher difficulties, 10 or 15%. The movement bonus makes up for her lack of foresight and the command penalty negation makes up for her poor leader management, a combat bonus would help make up for her sub-par divisions.

[Mickey3D]

Can agents be created to use Priority for a unit (Division) to have not just a certain amount of hits, but to also have specific sub-units, i.e. a sharpshooter, line infantry or all artillery?

I don't know very well modding but I'm afraid this is not possible.

[ArmChairGeneral]

Is it not possible, or just not possible with the currently available Agent commands?

The AGEWiki page linked above made it sound like more commands might be added if people were going to use them....

I would like to be able to adjust the propensity for attracting/absorbing elite elements, as well as priorities for support units like HQs and hospitals. For example, I could envision a Union agent scripted to try to capture Memphis via Island 10 wanting to equip itself with pontoons or marines.

[Gray Fox]

SkipHealth and SkipCohesion seem able to focus on those values of an element. Perhaps a similar agent might target a high value of discipline (line infantry), initiative (sharpshooter) or range (artillery) to scrutinize a sub-unit for prioritizing.

[ArmChairGeneral]

IT seems like, as written, it would be possible to write a combination of regular and agent AI scripts that could beef up capital defenses, potentially giving the AI better play against a DC rush. From what I can tell, you would need to hang all of the instructions/propensities on a specific general who would then be very likely to go to and stay in DC and then absorb available hits in the region into his stack.

THE documentation on the AGEWiki doesn't have complete instructions for AI modding, (it is more focused on scenario creation) so not sure how to actually do this in terms of what format to write the scripts in, where in the file structure to place the scripts, and how to actually go about attaching the agent script to the leader. It looks like the leader model would need to be modified either to attach the relevant script, or to adjust his preference for stacks of a certain composition, but the documentation is not clear on the exact steps for this.

[plasticpanzers]

Would it be possible to code the game for larger CSA corps and smaller US Corps and perhaps more promotions via time/event to allow them
to be built? This might allow for more corps to be built with better commanders. The current seniority system has little to do with favorites
of senior commanders given command of Corps, ability of commanders, or actual real seniority based upon time served in the Army and their
commissioning dates.

[ArmChairGeneral]

Looks like the short answer is yes, but only loosely (bearing in mind that I am only just getting started in this area and cannot tell you how to actually go about doing this yet).

You can adjust (I think in the model file?) a leader's propensity to be attracted to and want to command stacks with varying proportions of unit types. Not entirely sure how fine grained this is, but according to the wiki you can at least adjust by family. Using an agent script, you can then tell a leader how big of a priority his mission is compared to others, and how likely he is to absorb/be absorbed by other stacks in the region he is in. If you decreased all of your corps commanders absorb preference slightly, but made them all equal to each other, you could at least reduce the likelihood they would join together, and lower their preference for absorbing loose stacks in their region, hopefully keeping them a little smaller. You would need to attach the script to every 2* model (again, I think this is how it works).

I am sure there are more considerations, that it might be a tediously difficult task, and that you would not get exactly the results you are looking for, but on first glance something like this looks doable.

Promotions via event are straightforward. You could easily add an event that promoted a couple of leaders. There are one or two of these events in the April 61 scenario already that you could just cut and paste with different leaders and dates. Adding an event means you would create a new version of the scenario. This is compared to altering the AI scripts, which (again, I may have this wrong) would affect the behavior of the AI in all scenarios.

[Mickey3D]

May be I should have post this link that is the main page for modding.

[ArmChairGeneral]

Thanks Mickey3D. I looked through all those pages, but I am looking for a little bit more handholding, along the lines of "Create this type of file, set these parameters, save it as this kind of file in this spot and then do some other thing."

I opened the AI files in the database, and found the default agent scripts, and am messing around with csv splitter, etc., but am not sure exactly what the changes I am making are doing. I can understand all the scripting commands and the effect of the various database parameters. What is not on those wiki pages are exactly what steps need to be taken to enter and then activate a new script, and how exactly a new script will then be used/chosen by the engine.

(Also, I recognize that this is off the original topic, so when I have a sec I will open a new thread in the modding forum.)

[veji1]

Guys, my point wasn't even about optimal force construction, it was about just "some" force construction. I really think there should be a big big big cheat code for the AI in the winter pause in terms of force construction. I personnally don't think what Athena needs are combat bonuses, this just ruins the game because it makes battle numbers irrelevant. But some massive help for force construction would be a big bonus and easier (I think) than trying to script tons of agents.

For example a couple of days ago I played a july 61 game as CSA and in the 3 turns of late december to late january I took control of both sides :
- I funnelled many troops for the Union to Baltimore where I set up divisions and corps with good generals (one of the problems is that many good 2* generals are stuck commanding divisions), I cleaned up the mess in Washington and as McClellan was in Harpers Ferry, I sent him anothe corps. So basically in the east you have 1 army in Baltimore with 11 (!) Divisions set up in 3 corps and 1 army counter ; 2 corps with 4 divisions in Harpers Ferry (part of that same army) and in Washington another army and 2 corps with 8 Divisions overall. before I did this the Union had basically half as many divisions, 5 or 6 being strong ones and 5 or 6 being incomplete shells. The rest was a rump of overloaded stacks with massive penalties.
- As the Union had activated Kentucky without doing anything about it though I sent many troops from Ohio/Indiana, etc to Cincinnati (2 division's worth getting assembled), Louisville (3 weakish divisions under Grant/Halleck) and Lexington (a bit more than one division under Thomas).
- Finally I funnelled quite some troops in St Louis as well.
- Last I sent a division to Ashtrand (sp?) the province at the corner between Kentucky and WV from which one can set up an offensive in WV or to Prestonburg.

I also built up quite a few supply carts for the stacks outside the east because whether the east is full of supply carts, the other stacks were dangerously short.

Well let me tell you, I haven't had the chance to play since then (lots of work), but as I left the Union to its own devices in february I was shocked out how many troops it had ! With my CSA army in Virginia numbering something like 7 (big) divisions + one division at Williamsburg against Butler and one in Suffolk against an attempted invasion of Norfolk, I felt stretched thin. And with my 2 divisions in Missouri, one division under Polk in west Tennessee and 4 divisions under Johnston in Clarcksville (just south W of Bowling Green), I felt that come spring just a brutish push from the AI could be really painful. No need for the AI to be clever really : as the Union an AI (provided she has activation bonuses, a key element) that structures its forces well enough should be able to provide a good challenge. Sure it won't carve through the CSA, but the CSA player can expect to find himself defending the Culperer/Frederiscksburg axis fairly quickly or risk getting trashed.

This to me is the main problem.

This is why I would advocate a cheat giving the ability to the AI to basically instantly redeploy troops on the map during the winter months. With the same rules as the leaders (uninterrupted raillines) but for free, this would cheat would give the AI set up cities where she would concentrate : Baltimore, Cincinnati, St Louis being the obvious ones for the winter of 61/62. for the following winters the list would have to be expanded so that if such and such city is owned, it becomes the concentration point (say Nashville, Memphis, Charlotteville, etc...).

This would really help. Playing the CSA against the Union AI should be the easiest way to get a somewhat challenging game because the AI has a luxury of forces. Provided the AI has a bit of help in the Activation department mainly it should be able to move with organised forces. A fist made of 1 army and 3 corps of 11 division overall like the one I had assembled in Baltimore should just go through Alexandria to Manassas and go for battle there. Even if it losses, it will stil hurt the CSA massively because the CSA has a hard time recuperating from losses. As a player what I hate is that since the AI is badly structured, it has poor odds, and because it sort of evaluates its chances of success (ennemy detection, etc) before making a move, well it just doesn't move because it feels it can hurt me. And in the meanwhile with very very few losses my forces keep on growing and those odds become even worse.. but as soon as there are a few 8 000 losses battles, the CSA armies become so so brittle, you can't build anything more because all the conscripts go into replacement chits !


Honestly just loooking at the force I could assemble in just 3 turns of AI assisting in the dead turns of winter made me realise how much of a problem this is. Really the AI badly needs a hand for that : don't give the laser guns or super fast legs once in the campaigning fiels, but do give them infinite teleporting of troops and leaders to assemble more coherent forces behind the scene !

[veji1]

double post