So I have carried on with my experiments / having assisted the AI in setting up its forces better in St Louis, Louisville/Lexington, and the east coast with a well structured and very strong army of the Potomac, I eagerly waited for the year 62 to unfold :
- In the Transmississippi, some action if not much : Fremont (with Lyon as a division commander) attacked towards Springfield, bypassing Rolla and Lexington. After losing Springfield I took it again and the Union went back to St Louis which I didn't have enough forces to take, so after a small battle I went back to spreading my forces between Rolla/Lexington to deal with supply issues and that was that.
- In the West, tragically NOTHING happened. And i am playing with normal aggressive AI, Leutenant, + 1 detection and +2 activation. This is very frustrating. I understand that the combined operations of the sort Grand carried are just too complicated for the AI, but one could at least expect to see it bludgeon towards Bowling Green and take over Kentucky if the CSA doesn't have enough forces. even with the equivalent of 4 divisions in the Louisville/Lexington area, they just didn't move an inch.
- In the east at first nothing happened. I realised that the problem is that the AI just doesn't know what to do if the CSA is properly set up, controling Harpers Ferry/Leesburg/Manassas. It just stays there. I had a bit of fun with Lee attacking towards Gettysburg and then sending him and Jackson south of the Rappahanock (vacating Manassas) leaving just Longstreet in Harpers ferry. Suddenly this unclogged the AI like if it had been an old toilet : A 3 pronged attack developped with Butler and 3 very strong divisions taking Norfolk and Suffolk from Fort Monroe (forcing me to bring back Beauregard from Charleston and to rail 2 divisions from Tennesse + Buell and eventually 70 000 men crashing into Longstreet in Harpers Ferry and McDowell attacking frontally at Fredericksburg. I didn't end that weel for the AI of course, but at least I had some action : Butler just evaporated once I had Beauregard and 2 and1/2 divs in front of Suffolk, just going back to Washington as i later found out. McDowell just completely trashed himself on Lee/MTSGing Jackson losing 35 000 out of his 50 000 in a series of suicidal attacks on Fredericksburg and Buell defeated Longstree with moderate losses for Longstreet (5000 out of his 26 000 troops) but massive disruption meaning he hardly had any cohesion left. I had to send Jackson via train to Winchester to cover Longstreet's retreat all the way down the Shenandoah so that he could recuperate.
So there was some action in the east at least even if the end result is a USA with a NM in the doldrums (76) because of the massive losses incurred at Fredericksburg. But I must say it still leaves me frustrated because when in december 62 I loaded the Union side I could see that basically it had again made a massive mess of its forces in the east with the new Potomac army under Grant but all corps under Butler (who had just come back from Norfolk!!) AND hardly anything organised in the west.
Really the west seems to be the biggest weakness in the force prioritisation of the AI which is very problematic because the CSA player can than build up his troops in the east, never having to worry about what is happening in the west since nothing happens. Again I just don't understand it. the Union should be scripted to organise strong forces from Cincinnati, New Albany to take over the Louisville/Lexington area and from there march onto Bowling Green and to the Cumberland. This isn't too complicated, it is about allocating forces and building them up. Again I can see that one shouldn't hope for the AI to manage combined operation like Grant did, and that in this sense the whole confluent area of western Kentucky is basically unmanageable for the AI. But a central Kentucky offensive just isn't hard to do : Grant/Sherman + 4/5 divisions and lots of chariots and off you go to Bowling Green.
I honestly am very frustrated because I can see that even if I spend 4/5 turns playing both sides, cleaning up the big mess that is the east front and then building up the US troops in the west, it just won't move and try to invade south outside of the East. Very disheartening. I can dangle weakish troops in front of the AI, not organise my troops in the best way in the east so that I can role play ( Polk with 2 divisions is watching Columbus, far from Johnston and his now 2 divisions in Bowling Green) yet the AI won't do squat, even once has given it some force structure...
The way I see in terms of force structures, the AI really need to have series of checks to do a few times a year to adjust it's force structure.
- For example there is the problem that most of the good 2* (or even 3*) generals are still stuck commanding divisions in 62 (Burnside, Sedgewick, Kearny, Grant, Sherman, Lyons). The AI Should have checks to transfer those divisions to 1* leaders as soon as it can and when those leaders get their automatic promotion via event, one or better two 1* leaders should automatically appear by their side so that the division is instantaneously transfered to the lower ranked officer.
- To help it do that the creation of new generals should be spread out a lot more : there is this glut of officers in Washington and the AI jut can't manage to relocate some to the other fronts. Instead of the massive yearly apparation of leaders in Washington, apparitions should be spread out in say 2 events (january and mid-year) + during the automatic promotion events (see above) and be spread out in a few "hub" cities such as Cincinnati, St Louis and Baltimore, rather than Washington which today works like a blackhole. This would also help with the whole building of troops process by organising better the circulation and building up of divisions. Basically there should be a funnel system to help the AI instead of the Washington paralysis situation. Help the AI send the troop it builds to those 3 locations (say 15% to St Louis, 30% Cincinnati, 55% Baltimore) wher the leaders would appear, build up the divisions and then be sent (or even teleported if connected via railline, this is a cheat I would perfectly accept) to the big armies and corps.
There really needs to be progress made on this force structure issue for the AI, it is key for the replayability of the game and I must say I am disappointed that NO progress has been made on that matter since AACW.
- For example there is the problem that most of the good 2* (or even 3*) generals are still stuck commanding divisions in 62 (Burnside, Sedgewick, Kearny, Grant, Sherman, Lyons). The AI Should have checks to transfer those divisions to 1* leaders as soon as it can and when those leaders get their automatic promotion via event, one or better two 1* leaders should automatically appear by their side so that the division is instantaneously transfered to the lower ranked officer.
- To help it do that the creation of new generals should be spread out a lot more : there is this glut of officers in Washington and the AI jut can't manage to relocate some to the other fronts. Instead of the massive yearly apparation of leaders in Washington, apparitions should be spread out in say 2 events (january and mid-year) + during the automatic promotion events (see above) and be spread out in a few "hub" cities such as Cincinnati, St Louis and Baltimore, rather than Washington which today works like a blackhole. This would also help with the whole building of troops process by organising better the circulation and building up of divisions. Basically there should be a funnel system to help the AI instead of the Washington paralysis situation. Help the AI send the troop it builds to those 3 locations (say 15% to St Louis, 30% Cincinnati, 55% Baltimore) wher the leaders would appear, build up the divisions and then be sent (or even teleported if connected via railline, this is a cheat I would perfectly accept) to the big armies and corps.
There really needs to be progress made on this force structure issue for the AI, it is key for the replayability of the game and I must say I am disappointed that NO progress has been made on that matter since AACW.
-
ArmChairGeneral
- AGEod Grognard
- Posts: 997
- Joined: Thu Nov 21, 2013 9:00 am
- Location: Austin, TX, USA
One of the things I am interested in accomplishing with AI / agent scripting is to somehow get the Union to build up in Cincy and head to Memphis via Bowling Green and Donnelson. As it stands now, I have no problems as the CSA shooting up the L&N railroad from Nashville to Louisville the first turns KY lights up, and then threatening all of Indiana/Ohio from the juicy forward build points in N. Kentucky. The AI rarely does much in the way of offensive TN operations, although once I am threatening Ohio/Indiana she does eventually start to try to work in division sized forces behind me. I suspect part of the problem is that she can't "see" across KY while it is still neutral, so is unprepared for a CSA buildup in Nashville ready to quickly strike north.
The problem with organizing her forces by hand is that she jsut goes and messes them all up as soon as her situation changes. I am not sure how she decides how to organize stacks, (scriptedt, hardcoded, etc.) but I agree she could use some help overall. If any changes can get made here, she could use some help spreading out her support units too: no sense having three balloons in the DC stack, for example.
The problem with organizing her forces by hand is that she jsut goes and messes them all up as soon as her situation changes. I am not sure how she decides how to organize stacks, (scriptedt, hardcoded, etc.) but I agree she could use some help overall. If any changes can get made here, she could use some help spreading out her support units too: no sense having three balloons in the DC stack, for example.
Gray Fox wrote:I've just got to ask, "So you won the game?"
Athena doesn't have to attack a lot. Getting to 185 NM is only one victory condition. However, if she has 3001 VPs and you have 3000 in 1866, then her chaotic mess strategy wins.
GRayfox, let's say that it appears we have vastly different perspectives on what we look for in a game. You look at it in terms of winning, I personally have more fun if I feel immersed in the history of the time.
I don't care about winning or rather I know I can trash the AI anytime I want to if I want to. My point here is that I would like the AI to be set up in a way that provides a better gaming experience mainly by better organising its forces so that it can then be a bit more active.
ArmChairGeneral wrote:One of the things I am interested in accomplishing with AI / agent scripting is to somehow get the Union to build up in Cincy and head to Memphis via Bowling Green and Donnelson. As it stands now, I have no problems as the CSA shooting up the L&N railroad from Nashville to Louisville the first turns KY lights up, and then threatening all of Indiana/Ohio from the juicy forward build points in N. Kentucky. The AI rarely does much in the way of offensive TN operations, although once I am threatening Ohio/Indiana she does eventually start to try to work in division sized forces behind me. I suspect part of the problem is that she can't "see" across KY while it is still neutral, so is unprepared for a CSA buildup in Nashville ready to quickly strike north.
The problem with organizing her forces by hand is that she jsut goes and messes them all up as soon as her situation changes. I am not sure how she decides how to organize stacks, (scriptedt, hardcoded, etc.) but I agree she could use some help overall. If any changes can get made here, she could use some help spreading out her support units too: no sense having three balloons in the DC stack, for example.
Totally agreed. The AI Should build up in Cincinnati and New Albany from the start of the game, make them the staging points for its western offensive. There is a real issue here of force repartition with the AI overwhelmingly favoring the east, building some forces for Missouri, and leaving basically an empty space between the Missouri confluent and West Virginia.. this just doesn't make sense. the AI needs to have enough forces in the theatre to threaten the CSA. If it doesn't than the CSA can either run amok or build an impenetrable wall in Virginia with all those Tennesse/Alabama/mississippi divisions that are of no use in the west.
Personally I think it is a bit much to say that the AI should build up in certain areas. That is too deterministic for me. But I heartily agree that it should put together better forces, and that it should have more possibility to use its better generals better.
Regarding the force build ups, I understand that an AI is unlikely to be able to take the 1000 small facts into consideration and weigh them as a human does, so there does need to be some guidance. I've always favored the idea that the AI have several "Grand Plans" for each region, and by randomly selecting one to provide a different, but reasonably active, game each time. Sure if you play endlessly you could eventually figure out that a concentration in Springfield followed by a move to X means Y will happen later, but until true AI occurs I don't see how to totally avoid that. The scripting could always include the possibility of the plan being changed later as well.
But I think all that starts with better (not necessarily optimal min/max style [IMHO]) force design and leader usage.
Regarding the force build ups, I understand that an AI is unlikely to be able to take the 1000 small facts into consideration and weigh them as a human does, so there does need to be some guidance. I've always favored the idea that the AI have several "Grand Plans" for each region, and by randomly selecting one to provide a different, but reasonably active, game each time. Sure if you play endlessly you could eventually figure out that a concentration in Springfield followed by a move to X means Y will happen later, but until true AI occurs I don't see how to totally avoid that. The scripting could always include the possibility of the plan being changed later as well.
But I think all that starts with better (not necessarily optimal min/max style [IMHO]) force design and leader usage.
-
ArmChairGeneral
- AGEod Grognard
- Posts: 997
- Joined: Thu Nov 21, 2013 9:00 am
- Location: Austin, TX, USA
RickInVA,
Fortunately (according to my current understanding) the agent scripts are available choices for the AI, not ironclad instructions. So, if for example, I write a script for invading TN via Bowling Green, and one for invading TN amphibiously at Paducah, then the AI will pick between them, and also potentially not do either. The propensities can be adjusted to increase or decrease the likelihood of any given script being chosen, but from game to game the AI will not always follow the same path. The agent scripts allow you to put some extra arrows in its quiver without guaranteeing that it will do the same thing every time.
It is perfectly normal in most game titles for the AI to be adjusted to counter the most common strategies and tactics players come up with as patches and updates are released. While I don't think that the AI should be FORCED to build up in certain locations, the map has a logic of its own. Cincy is a big production center, sits atop a large force pool, and is smack dab in the middle of the largest concentration of strategic and objective cities on the board. It makes sense for the AI to focus on building there, especially since if it doesn't it leaves a ton of easily available NM and VPs available for the CSA to get their hands on. If most human players find that they tend to build up in Cincy (or whatever city makes sense), it makes sense that the AI should at least be equipped to consider it as well.
For that matter, since Fox and others have demonstrated that a capital rush is an effective technique, it would be cool to see the AI try it herself from time to time.
Fortunately (according to my current understanding) the agent scripts are available choices for the AI, not ironclad instructions. So, if for example, I write a script for invading TN via Bowling Green, and one for invading TN amphibiously at Paducah, then the AI will pick between them, and also potentially not do either. The propensities can be adjusted to increase or decrease the likelihood of any given script being chosen, but from game to game the AI will not always follow the same path. The agent scripts allow you to put some extra arrows in its quiver without guaranteeing that it will do the same thing every time.
It is perfectly normal in most game titles for the AI to be adjusted to counter the most common strategies and tactics players come up with as patches and updates are released. While I don't think that the AI should be FORCED to build up in certain locations, the map has a logic of its own. Cincy is a big production center, sits atop a large force pool, and is smack dab in the middle of the largest concentration of strategic and objective cities on the board. It makes sense for the AI to focus on building there, especially since if it doesn't it leaves a ton of easily available NM and VPs available for the CSA to get their hands on. If most human players find that they tend to build up in Cincy (or whatever city makes sense), it makes sense that the AI should at least be equipped to consider it as well.
For that matter, since Fox and others have demonstrated that a capital rush is an effective technique, it would be cool to see the AI try it herself from time to time.
In the 20th century a computer could beat the best checkers player. In this century, one has already beat the best chess master. All too soon we'll see when players will be asking for easier PC game AI. Perhaps it would be possible to see a small decision tree of actual Athena agents to get an idea of what is being scripted and what it looks like.
I'm the 51st shade of gray. Eat, pray, Charge!
Gray Fox wrote:Perhaps it would be possible to see a small decision tree of actual Athena agents to get an idea of what is being scripted and what it looks like.
Just my 2 cents, but I hope not as I sure don't want to know ahead of time what is in my opponent's mind. The fun of the game would be diminished slightly if I somehow knew what my opponent might do next. And I'm sure someone would do some spreadsheet calculations on the odds that X, Y, or Z would happen. I'd rather the AI be somewhat of a mystery.
Thanks guys for your different comments.
I have pursued my adventures, starting a new april 61 game as the CSA but taking over the Union as well in early november 61 until early april 62. This a very long time span but it has allowed me to funnel the troop build up in key places and wait for the authorisation of corps so that I could complete the force structure and see what would happen in 62. I should say that there had been a fair bit of fighting in summer of 61 in Missouri ending with poor Fremont and Lyons having gone too far and been defeated in front of Springfield by the arriving McCullouch and having to retreat poorly north from which I extracted them to the north (St Joseph I think, the town with the port) and sent them by river to St Louis to recuperate, and one silly try by the AI to just land the Sigel and Sturgis brigades on top of memphis ending in those brigades sunk by the guns of island 10 and fort pillow... But nothing happened in the west and nothing much in the east either, just a bit of jostling for Harpers Ferry between Banks/Butler and Johnston, ending with Johnston in Winchester (more by choice than anything) and the Union in Harpers.
Now before I delve into what I did to help the AI, let me just state my "game philosophy". I am not saying it is superior to anyone's or anything, just explaining it so that other members understand where I come from : I know I won't get a truly competitive game from the AI, but if I put limitations on my "hindsight" and sort of "role play" constraints, I can give the AI more of a fighting chance to give me some action and fun. For example as the CSA, to emulate the fact that it lacked a strong unified central command and the strong autonomy of the states, I always build all the militia I can (in pairs so that I form those 2 militia regiments brigades) in all state and leave them in the state. They are not allowed to leave the state except when difficult circumstances require reinforcements. So I don't build up many super beautiful divisions with lots of nifty optimized brigades. Sure I arrange the divisions as best as I can (sharpshooters, etc) and build some regular troops as well when I have cash to spare, but the result is that my armies are smaller and when engaged in hard campaigning, most of the reinforcement they get in the first weeks are militia. That way my armies are less intimidating to the AI I guess and I get more fun because I have to think "can AS Johnson hold south Kentucky with just 4 divisions of 15ish elements and a good third of militia (Tennessee and Kentucky militiamen mainly) ?"
Now that I have stated this, this is what I did to help the AI :
-In the Far west, nothing much, haven't built troops for the Union (nor the CSA actually) and just left it like it is with the beginning troops.
-In Missouri I have funneled some troops from states like Iowa, Wisonsin and local Missouri troops to St Louis to give the Union good defensive capabilities, but nothing much more there, particularly because this is a place where I have noted the AI is often active but sort of wastes its troops away acting like the ball in the pinball between Lexington/Rolla/Springfield. What I have done though is give it sensible command structure : one army under Fremont but it is an empty counter and a corps under Lyon with the troops, around 1000cp.
-for a Mississippi offensive I have built many Illinois troops in Cairo and funneled quite a bit fromt he north so that I have built up there Pope, Summer and Dix and about 5 one* generals into 2 good divisions and 1 weakish one in a troop based at Charleston (with 3 supply wagons to boot). It is a bit militia heavy because the AI had built and sent quite a bit of militia there. The idea is to see what the AI would do there : going for Poplar's Bluff and then inland Missouri, going for Columbus and west Kentucky, going south towards New Madrid, etc... All in all something like 1500cp but unfortunately it isn't a corps per se as neither Fremont nor Buell's radius could reach it.
-for a Kentucky campaign I have massively proped up the AI and built a very strong force : 1/ the Army of Tennessee under Buell when he appeared in Evansville with 2 corps, one led by Grant and one led by "forgot who but decent too". Grant's corps has 3 strong divisions and the other one 2, the army is an empty shell. I built up many Indiana troops and sent lots of Michigan and Ohio troops there as well. Again I made sure they have one supply wagon per division. All in all I have a bit more than 2500cp there. 2/ The army of Ohio or whatever its name built in Louisville around Halleck with a corps under Whipple and a corps under "I forgot who" in Lexington. I funnelled troops via Cincinnati and the raillines and have built an army of a bit less than 2500cp with 5 divisions, 3 under Whipple in Louisville and 2 under the other corps starting in Lexington. So all in all the Union has about 5000cps for a kentucky campaign. I should state that i bought the "kentucky intervention chit" for the AI before letting it go.
-In West virginia McClellan is still fooling around with one strong division (500cps) about to at last take Clarksburg when the whether allows it.
-In the east I just funnelled the troops the AI had built (but hardly built any myself, there was plenty enough) into sensible divisions and then divided those into sensible corps : one army under Butler in Harpers Ferry with 2 corps of about 3 divisions each and 2 divisions for the AI. It actually absorbed Banks army there and Banks is now a corps commander. Overall it has around 4000cp. And in Alexandria I stationned the massive army under McDowell with 3 corps of 4 divisions + 3 divisions in the army : it has close a bit more than 8000cp.
Of course one last point, I spent a tedious amount of time redeploying officers around to ensure that the StLouis, Cairo/Charleston, Evansville, Louisville/Lexington and Harpers Ferry funnels had enough officers to properly build.
This is what I did for the Union and the situation I left it with when I switched back to playing exclusively the CSA in late april 62. My CSA dispositions were as follow :
- Nothing new in the Far West
- In Missouri/IT I have Watie's indian division (300cp) fooling around in the stockade area, a cavalry division under Shelby (400cp) sitting at Lexington and 2 divisions with lots of militia under McCullouch and Price at Rolla (about 900cp). no good C&C since there is no general higher than 1*.
- In the Mississippi area there is Edward Johnson and that long locked Arkansas army complemented by the Missouri brigade, one Arkansas brigade with arty, a couple of sharpshooters and 1 supply wagon (god they are expensive when you play the CSA). this troop is waiting to be unlocked and has about 900/1000cps and transformed into 3 middling divisions under Hardee and 2 others I can't remember. Add to this that Polk is standing at Humboldt with about 800cps (one good division under Bragg, one smallish under Anderson). As soon as Lee got freed, I sent Johnston there to take command of the Army of Mississippi, taking over Polk's stack.
- In Kentucky AS Johnson is standing at Bowling Green with 1800cps and 4 divisions of average quality.
- In west Virginia I built 2 weakish divisions under Floyd and Ewell and put Beauregard as army chief (Army of the Shenandoah). Some 700cps with 2 supply wagons and a depot built at Lexington (just east of Covington), could have left all that empty but figured I have some fun playing a WV campaign.
- In Virginia there is one corps under Longstreet with 3 strong divisions in Winchester (1600cps) facing the 4000cps of Bulter's army, 2200cps under Lee in Manassas (I think 4 strong divisions) and 1300cps under Jackson in Falmouth (3 average divisions) + Stuart and Thompson with small cavalry divisions screening the gaps. So you basically have Lee and Jackson, ie a bit less than 4000cps facing 8000cps.
Now remember that I have quite some troops in all the states, as militia mainly but with the original contingents : Wilder in Charleston and Evans in Savannah each have a strong division of 500cps. Bushrod Johnson is standing at Williamsburg with a 350cp division and Huger is in Norfolk with a pathetic 200cp division. In Mobile and New Orleans there are weakish Militia divisions of 300cps and there is a fair bit of militia 2 regiment brigades spread around the board. The Idea is that if really pressed I can call either those divisions, or part of their troops, or some militia brigades and rail them to my armies. Sort of emulates how the governors kept many forces for themselves but could be strongarmed into helping a bit if really needed.
So now to the game. What happened when I gave the AI full control of the Union in early May 62 ? Well we are now in July 62 and here we are.
- in the Far west nothing, we are both content just ignoring this.
- in Missouri, the AI probably has sensed the rapport the force and isn't moving towards Rolla, but it has sent Fremont and Lyons towards Charleston/Columbus !
- In Mississippi : probably reinforced by one of the corps in Evansville, the AI has attacked Paducah and Columbus and has 2000cps standing in in Columbus, and I think one division in New Madrid. Johston and his 900cps under Polk await the onslaught, as soon as the Arkansas army frees up, I send one division there (with 450ish cps) and leave Johson with just 600cps in his corps on the west side of the Mississippi. As Johnston and his now 1400ish cps ponder what to do and try to get to see what is in from of him using Forrest's cavalry division, the Union troops suddenly vaporise ??? Where did they go, no clue... My big big big fear is that some of it is travelling by train or something all the way to the east....
- in Kentucky the double pronged force structure from Evansville and Louisville works like a charm (by AI standards of course) : Halleck and his 2000ish Cp take Munfordsville and threaten Bowling Green, as a result AS Johson crosses the river and they meet for battle with about 30 000 soldiers each. somewhat of a draw with 4000 men lost on both sides. But the other prong is approaching from the north west under some corps (not Grant, I think Pope, who was sent from Cairo ??) and as AS goes back to Bowling Green to meet the threat, another battle with 25000 men and another 4000 lost. This happens a couple of times and as we reach july I have had to abandon Kentucky with the Army of Tennesse and cross the river bakc to Nashville before risking to lose my troops. in Nashville await desperatly needed reinforcements : Evans division sent from Savannah and some gaggle of Alabama/Georgia/Louisiana militia + a couple of sharpshooting brigades to beef up the weary Army of Tennessee... THIS IS BY FAR THE MOST FUN I HAVE HAD IN THE WEST AGAINST THE AI! Now I face the decision of having to send troops for Johnston army of the Mississippi to the army of the Tennesse which means that the Mississippi will be quite vulnerable...
- West Virginia, Beauregard's offensive is an unmitigated disaster : He meets McClellan's troops (one well entrenched division) east just in the hilly province between Clarcksburg/Lewisburg//Parkersburg triangle and loses and just can't seem to be able to retreat properly, and now there is mud and he is crawling and the cohesion is fading and the supplies are dwindling... A shell of an army is goint to come back to Virginia I fear.. Frustrating but enjoyable as well from an rp perspective.
- Virginia... Ah Virginia... The pure complete and utter stalemate. it seems that I have basically blocked the game even more than usual because the Union is too strong and well structured for me to really try something (I thought about sending all 3 corps to Harpers Ferry to take on the 4000 Butler stack (I have noticed that Grant appears to be commanding a corps there, so he has been sent east), but I would have left the road to Richmond completely open and with something lik 5500cps at the very best, my odds weren't great. And with just Lee and Longstreet, with 4000cps it wasn't going to cut it. But the AI itself seems content to just sit there : it has retreated McDowell's army and corps from Alexandria to Washington so there is this humongous 10 000cps blob in Washington now, not doing squat.. As usual Virginia sucks. I know what I could do to make the situation more fluid : retreat down the Shenandoah and to Fredreicksburg /Albermale / Culperer : it works everyime, suddenly the AI starts moving. It really seems that when the front isn't deep enough the AI just doesn't know what to do...
- No seabord operation whatsoever : to bad because I had concentrated the northern fleet in Annapolis and Boston, all properly set up and stuff, all the AI would have had to do is drop one of the McDowell's corps in the boats or something and it would have had a great 3 or 4 divisions with plenty of wagons invasion force.. Shame because it would have been fun to watch my militia divisions trying to face the threat and forcing me to divert some troops.
This is where we are in late july 62 : to some extent I have seen how setting up the force structure of the AI can help it, as well as its lasting limitations... i'll carry on till the end of 62 and if I am brave enough, I'll do the same thing to set up 63, but I fear that the further you go from the starting positions, the messier it becomes for the AI : for example will it think to build a depot in Bowling Green ? not sure and without it it will starve...
I'll keep you posted.
I have pursued my adventures, starting a new april 61 game as the CSA but taking over the Union as well in early november 61 until early april 62. This a very long time span but it has allowed me to funnel the troop build up in key places and wait for the authorisation of corps so that I could complete the force structure and see what would happen in 62. I should say that there had been a fair bit of fighting in summer of 61 in Missouri ending with poor Fremont and Lyons having gone too far and been defeated in front of Springfield by the arriving McCullouch and having to retreat poorly north from which I extracted them to the north (St Joseph I think, the town with the port) and sent them by river to St Louis to recuperate, and one silly try by the AI to just land the Sigel and Sturgis brigades on top of memphis ending in those brigades sunk by the guns of island 10 and fort pillow... But nothing happened in the west and nothing much in the east either, just a bit of jostling for Harpers Ferry between Banks/Butler and Johnston, ending with Johnston in Winchester (more by choice than anything) and the Union in Harpers.
Now before I delve into what I did to help the AI, let me just state my "game philosophy". I am not saying it is superior to anyone's or anything, just explaining it so that other members understand where I come from : I know I won't get a truly competitive game from the AI, but if I put limitations on my "hindsight" and sort of "role play" constraints, I can give the AI more of a fighting chance to give me some action and fun. For example as the CSA, to emulate the fact that it lacked a strong unified central command and the strong autonomy of the states, I always build all the militia I can (in pairs so that I form those 2 militia regiments brigades) in all state and leave them in the state. They are not allowed to leave the state except when difficult circumstances require reinforcements. So I don't build up many super beautiful divisions with lots of nifty optimized brigades. Sure I arrange the divisions as best as I can (sharpshooters, etc) and build some regular troops as well when I have cash to spare, but the result is that my armies are smaller and when engaged in hard campaigning, most of the reinforcement they get in the first weeks are militia. That way my armies are less intimidating to the AI I guess and I get more fun because I have to think "can AS Johnson hold south Kentucky with just 4 divisions of 15ish elements and a good third of militia (Tennessee and Kentucky militiamen mainly) ?"
Now that I have stated this, this is what I did to help the AI :
-In the Far west, nothing much, haven't built troops for the Union (nor the CSA actually) and just left it like it is with the beginning troops.
-In Missouri I have funneled some troops from states like Iowa, Wisonsin and local Missouri troops to St Louis to give the Union good defensive capabilities, but nothing much more there, particularly because this is a place where I have noted the AI is often active but sort of wastes its troops away acting like the ball in the pinball between Lexington/Rolla/Springfield. What I have done though is give it sensible command structure : one army under Fremont but it is an empty counter and a corps under Lyon with the troops, around 1000cp.
-for a Mississippi offensive I have built many Illinois troops in Cairo and funneled quite a bit fromt he north so that I have built up there Pope, Summer and Dix and about 5 one* generals into 2 good divisions and 1 weakish one in a troop based at Charleston (with 3 supply wagons to boot). It is a bit militia heavy because the AI had built and sent quite a bit of militia there. The idea is to see what the AI would do there : going for Poplar's Bluff and then inland Missouri, going for Columbus and west Kentucky, going south towards New Madrid, etc... All in all something like 1500cp but unfortunately it isn't a corps per se as neither Fremont nor Buell's radius could reach it.
-for a Kentucky campaign I have massively proped up the AI and built a very strong force : 1/ the Army of Tennessee under Buell when he appeared in Evansville with 2 corps, one led by Grant and one led by "forgot who but decent too". Grant's corps has 3 strong divisions and the other one 2, the army is an empty shell. I built up many Indiana troops and sent lots of Michigan and Ohio troops there as well. Again I made sure they have one supply wagon per division. All in all I have a bit more than 2500cp there. 2/ The army of Ohio or whatever its name built in Louisville around Halleck with a corps under Whipple and a corps under "I forgot who" in Lexington. I funnelled troops via Cincinnati and the raillines and have built an army of a bit less than 2500cp with 5 divisions, 3 under Whipple in Louisville and 2 under the other corps starting in Lexington. So all in all the Union has about 5000cps for a kentucky campaign. I should state that i bought the "kentucky intervention chit" for the AI before letting it go.
-In West virginia McClellan is still fooling around with one strong division (500cps) about to at last take Clarksburg when the whether allows it.
-In the east I just funnelled the troops the AI had built (but hardly built any myself, there was plenty enough) into sensible divisions and then divided those into sensible corps : one army under Butler in Harpers Ferry with 2 corps of about 3 divisions each and 2 divisions for the AI. It actually absorbed Banks army there and Banks is now a corps commander. Overall it has around 4000cp. And in Alexandria I stationned the massive army under McDowell with 3 corps of 4 divisions + 3 divisions in the army : it has close a bit more than 8000cp.
Of course one last point, I spent a tedious amount of time redeploying officers around to ensure that the StLouis, Cairo/Charleston, Evansville, Louisville/Lexington and Harpers Ferry funnels had enough officers to properly build.
This is what I did for the Union and the situation I left it with when I switched back to playing exclusively the CSA in late april 62. My CSA dispositions were as follow :
- Nothing new in the Far West
- In Missouri/IT I have Watie's indian division (300cp) fooling around in the stockade area, a cavalry division under Shelby (400cp) sitting at Lexington and 2 divisions with lots of militia under McCullouch and Price at Rolla (about 900cp). no good C&C since there is no general higher than 1*.
- In the Mississippi area there is Edward Johnson and that long locked Arkansas army complemented by the Missouri brigade, one Arkansas brigade with arty, a couple of sharpshooters and 1 supply wagon (god they are expensive when you play the CSA). this troop is waiting to be unlocked and has about 900/1000cps and transformed into 3 middling divisions under Hardee and 2 others I can't remember. Add to this that Polk is standing at Humboldt with about 800cps (one good division under Bragg, one smallish under Anderson). As soon as Lee got freed, I sent Johnston there to take command of the Army of Mississippi, taking over Polk's stack.
- In Kentucky AS Johnson is standing at Bowling Green with 1800cps and 4 divisions of average quality.
- In west Virginia I built 2 weakish divisions under Floyd and Ewell and put Beauregard as army chief (Army of the Shenandoah). Some 700cps with 2 supply wagons and a depot built at Lexington (just east of Covington), could have left all that empty but figured I have some fun playing a WV campaign.
- In Virginia there is one corps under Longstreet with 3 strong divisions in Winchester (1600cps) facing the 4000cps of Bulter's army, 2200cps under Lee in Manassas (I think 4 strong divisions) and 1300cps under Jackson in Falmouth (3 average divisions) + Stuart and Thompson with small cavalry divisions screening the gaps. So you basically have Lee and Jackson, ie a bit less than 4000cps facing 8000cps.
Now remember that I have quite some troops in all the states, as militia mainly but with the original contingents : Wilder in Charleston and Evans in Savannah each have a strong division of 500cps. Bushrod Johnson is standing at Williamsburg with a 350cp division and Huger is in Norfolk with a pathetic 200cp division. In Mobile and New Orleans there are weakish Militia divisions of 300cps and there is a fair bit of militia 2 regiment brigades spread around the board. The Idea is that if really pressed I can call either those divisions, or part of their troops, or some militia brigades and rail them to my armies. Sort of emulates how the governors kept many forces for themselves but could be strongarmed into helping a bit if really needed.
So now to the game. What happened when I gave the AI full control of the Union in early May 62 ? Well we are now in July 62 and here we are.
- in the Far west nothing, we are both content just ignoring this.
- in Missouri, the AI probably has sensed the rapport the force and isn't moving towards Rolla, but it has sent Fremont and Lyons towards Charleston/Columbus !
- In Mississippi : probably reinforced by one of the corps in Evansville, the AI has attacked Paducah and Columbus and has 2000cps standing in in Columbus, and I think one division in New Madrid. Johston and his 900cps under Polk await the onslaught, as soon as the Arkansas army frees up, I send one division there (with 450ish cps) and leave Johson with just 600cps in his corps on the west side of the Mississippi. As Johnston and his now 1400ish cps ponder what to do and try to get to see what is in from of him using Forrest's cavalry division, the Union troops suddenly vaporise ??? Where did they go, no clue... My big big big fear is that some of it is travelling by train or something all the way to the east....
- in Kentucky the double pronged force structure from Evansville and Louisville works like a charm (by AI standards of course) : Halleck and his 2000ish Cp take Munfordsville and threaten Bowling Green, as a result AS Johson crosses the river and they meet for battle with about 30 000 soldiers each. somewhat of a draw with 4000 men lost on both sides. But the other prong is approaching from the north west under some corps (not Grant, I think Pope, who was sent from Cairo ??) and as AS goes back to Bowling Green to meet the threat, another battle with 25000 men and another 4000 lost. This happens a couple of times and as we reach july I have had to abandon Kentucky with the Army of Tennesse and cross the river bakc to Nashville before risking to lose my troops. in Nashville await desperatly needed reinforcements : Evans division sent from Savannah and some gaggle of Alabama/Georgia/Louisiana militia + a couple of sharpshooting brigades to beef up the weary Army of Tennessee... THIS IS BY FAR THE MOST FUN I HAVE HAD IN THE WEST AGAINST THE AI! Now I face the decision of having to send troops for Johnston army of the Mississippi to the army of the Tennesse which means that the Mississippi will be quite vulnerable...
- West Virginia, Beauregard's offensive is an unmitigated disaster : He meets McClellan's troops (one well entrenched division) east just in the hilly province between Clarcksburg/Lewisburg//Parkersburg triangle and loses and just can't seem to be able to retreat properly, and now there is mud and he is crawling and the cohesion is fading and the supplies are dwindling... A shell of an army is goint to come back to Virginia I fear.. Frustrating but enjoyable as well from an rp perspective.
- Virginia... Ah Virginia... The pure complete and utter stalemate. it seems that I have basically blocked the game even more than usual because the Union is too strong and well structured for me to really try something (I thought about sending all 3 corps to Harpers Ferry to take on the 4000 Butler stack (I have noticed that Grant appears to be commanding a corps there, so he has been sent east), but I would have left the road to Richmond completely open and with something lik 5500cps at the very best, my odds weren't great. And with just Lee and Longstreet, with 4000cps it wasn't going to cut it. But the AI itself seems content to just sit there : it has retreated McDowell's army and corps from Alexandria to Washington so there is this humongous 10 000cps blob in Washington now, not doing squat.. As usual Virginia sucks. I know what I could do to make the situation more fluid : retreat down the Shenandoah and to Fredreicksburg /Albermale / Culperer : it works everyime, suddenly the AI starts moving. It really seems that when the front isn't deep enough the AI just doesn't know what to do...
- No seabord operation whatsoever : to bad because I had concentrated the northern fleet in Annapolis and Boston, all properly set up and stuff, all the AI would have had to do is drop one of the McDowell's corps in the boats or something and it would have had a great 3 or 4 divisions with plenty of wagons invasion force.. Shame because it would have been fun to watch my militia divisions trying to face the threat and forcing me to divert some troops.
This is where we are in late july 62 : to some extent I have seen how setting up the force structure of the AI can help it, as well as its lasting limitations... i'll carry on till the end of 62 and if I am brave enough, I'll do the same thing to set up 63, but I fear that the further you go from the starting positions, the messier it becomes for the AI : for example will it think to build a depot in Bowling Green ? not sure and without it it will starve...
I'll keep you posted.
Who is online
Users browsing this forum: No registered users and 12 guests