Some Questions Playing the Union - 1861 Start

[jscott991]

So despite saying I was going to give up on this game, I've kept chugging along. There is so much I don't like about it (AGEOD's interface is just so clunky compared to the Paradox games I've grown used to), but I just have this feeling that if I could get over it, it's the best Civil War game out there.

I have some questions now that I'm trying to play more competently. My goal in any Civil War game is to refight the Peninsula Campaign on time so a lot of my gameplay is geared around that.

1. How are the Western Union armies supposed to be organized?

I sat around for a long time doing basically nothing in the West because I couldn't figure out how to move troops without losing tons of cohesion. I thought maybe getting Grant or some other army commander going would help. Eventually I gave up because of Grant's rank and promoted Halleck and Buell to army command and started crawling through western Tennessee and central Kentucky. It's June 1862 and I finally took Fort Donelson (or the westernmost one) and Bowling Green. This is well behind the historic pace, but even waiting this long, AS Johnston drove me back. There just aren't any troops in the West and there are even fewer commanders.

How do people generally set this area up? What's reasonable to accomplish by the end of 1861? Mid-1862?

2. Where are all the corps commanders?

It's an exciting moment when you can finally make corps. But there are almost no historic corps commanders available. Is this by design? In the east, the first four Union corps commanders should be McDowell, Sumner, Keyes, and Heintzelman. The next two are Franklin and Porter. From what I can tell, Keyes, Porter, and Heintzelman will never qualify. In fact, I found myself using C. Hamilton as a corps commander. Also, Whipple and Berry are major generals before Keyes and Heintzelman. Did I do something wrong or is there just something a little screwy about the Union's two star generals? If so, I will try to mod the files a bit to get Keyes, Porter, and Heintzelman to their proper ranks, but I wanted to make sure I wasn't missing anything.

3, Why does the CSA continually push north in Maryland and Pennsylvania, retreat, and then do it all over again?

I used to have a major problem in CW1 with the CSA using J. Johnston to try to get to Pittsburgh and his army just disintegrating on the way back. This happened all the time and crippled the CSA early. In CW2, Frederick seems to serve this purpose. The main CSA armies go back and forth invading Frederick and retreating into the Shenadoah. They don't defend Manassas and they've let me land on the Peninsula and get to right outside Richmond with basically nothing in front of me. Is this typical?

4. I don't understand cohesion and effective movement at all.

Whenever I move my troops, even just simple movements (like say back and forth between Washington and Frederick), they lose a ton of cohesion. The AI CSA doesn't seem to have this problem. Am I missing something about how to move or stack composition that is causing this? Also, why does it take so long to march between areas sometimes? I've had simple moves from one region of Kentucky to another or from Washington to Frederick take 3+ weeks. Again, am I missing something basic about movement?

Once you can form divisions and such the game becomes a lot more fun, but I still think I'm missing something basic. I feel like I'm moving in molasses to accomplish anything. It's so hard to get a good stack of troops in place (particularly out west) and then get it moving. Keeping up with the historic pace of the Union (which shouldn't be that hard) is just impossible.

Any tips would be great.

[Blood and Thunder Brigade]

Regarding corps commanders; Keyes will be automatically promoted around March or April of 1862 if I remember correctly.
Heintzelman and Porter are also promotable, but that'll come with the certainty of upsetting a more senior general. Depending upon who that general is I'll sometimes promote someone ahead of them anyway.

[jscott991]

How do you promote someone?

Also, I don't think Keyes was promoted in my game. He's still a brigadier in June of 1862.

[jscott991]

Just to bring home the CSA/no defending Richmond thing, what is going on here?

I don't understand what the Confederates are trying to accomplish with all of their forces so far from Richmond when I'm on top of it with three corps.

[pgr]

Regarding movement and cohesion, how much are you using RR movement? It is far and away the fastest way to go. If you are matching in safe areas, setting to green helps conserve cohesion. When attacking, weather makes a big difference. ( Mud) Keeping inside the CP limit helps as does having a medical support unit in the stack.

Regarding the West, 1861 is really tight for resources, so you have to manage your builds really well if you want be rolling by February 62. I tend to RR most of the free units East and split my builds between east and west. I also don't pick major fights in 61, because if you have to divert extra manpower into replacements early, it can stall your buildup. As for attacking in the West, use the river with the Navy. Aim for 3 divisions under Grant by Feb 62 that can be loaded on porter's fleet and can be landed someplace, and once you have it, funnel follow on forces there.

To promote leaders, they need to fight (specifically lead stacks/units that kill enemy elements) so select your protégé and set him up for an easy win. Atlantic fort busting is popular. Remember to storm the forts, because seniority comes from the battle not winning a seige. Btw, Grant starts very close to promotion, taking island no 10 or Ft Donelson by storm is usually sufficient.

The AI seems to be built to be agressive. If it sees VP locations "weakly" garded, it goes for them. Therefore, if you are playing Union, try lowering the AI detection and agression. (Or PBEM

[Teatime]

My comments, but it's been a while since I played just the base game so take these with some qualifiers .....

I found the key to getting the Union rolling is getting those volunteers/Militia/Conscripts upgraded to reg inf early. To do that you want your training officers (McClellan, Halleck & Sigel) running "training camps". McClellan you can get away with while he is running the AoP, just run all non upgraded through his command stack and let the corps do the fighting.

Build divisions in your camps and get them out into the field.

Hardest part is getting the arty to feed into those divisions, you do need to plan your builds

1. The West
I generally look to the historical timeline as a guide. I should have Nashville/Memphis/New Orleans by the end of 62, Chattanooga & Vicksburg by 63.

Rivers are your highways and protect them well yourself with forts.

Depots can be built with Flat Boats on harbours - Cheaper than Supply Trains

As mentioned. 3 divisions with Grant

Be ready for KY activation....
I find a division or 2 at Evansville allows you to riverine move down to grab Bowling Green (tidy up the railways after) and then drop a Flat Boat at Bowling Green for a depot.
Grab Paducah and be ready to build a fort and drop a Flat Boat for a depot as soon as you can

Also forts on St Louis and Cairo will protect those

4 arty elements + 4 supply train elements to build a fort

As a note, I find it tough to replicate the costal campaign as the Union doesn't seem to be able to spare the troops for taking those forts (and leaving a garrison) in 61/62.

2. The corps commanders
Yeh, you need to make an effort to get your good guys promoted .. pgr is spot on, so get your good upgradeable commanders into independent stacks where they get the victory credits .. assaulting forts is the way to go .. or catching those pesky cavalry raids though that can be an effort in frustration.

I don't think the game has the timeline quite right on when generals became 2* or who was available early.

3. I wish I knew that one but may be an issue with how tight the map is in the east as it's hard to do those flank moves across only 3 zones with the river lines to boot. I find Athena has a habit of sending big armies into West Virginia as well.

4. As mentioned, Mud is bad so look at the areas you are moving to, If it takes 9 days to move one region you are probably dealing with mud and it will drain your cohesion. Use riverine and rail when you can, though riverine in bad weather will hurt big time. Use transports including riverine where you can over strategic riverine as well.

Hope that helps

[Gray Fox]

Remember, this is a game, not a recreation. Have you read this:
viewtopic.php?f=331&t=43074

1. I form a "shield" of Divisions from St. Louis to Pittsburgh entrenched in the major cities along the rivers. This allows me to concentrate on the East. In your pic, D.C. is very weakly held. The AI won't hit you there like I would.

2. Take Grant right from the start as stack commander of a Division under Lyons and capture a coastal fort. That should allow them to be promoted to Army and Corps, respectively. Do this with other Generals as needed.

3. The game settings will make the AI more or less adventurous. You'll need to play with them to find a "sweet spot" that is to your liking. I play at Colonel difficulty, so you probably don't want my settings, just yet.

4. Does your stack have a hospital or HQ unit with it? These help recover cohesion.

Good luck!

[jscott991]

Remember, this is a game, not a recreation. Have you read this:
viewtopic.php?f=331&t=43074

1. I form a "shield" of Divisions from St. Louis to Pittsburgh entrenched in the major cities along the rivers. This allows me to concentrate on the East. In your pic, D.C. is very weakly held. The AI won't hit you there like I would.

Well, presumably you wouldn't have most of your CSA forces in West Virginia or deep in the Valley.

I'm going to have to start again. Taking Richmond in mid-1862 without fighting a real battle is no fun.

I have read the big tip thread. A lot of it is helpful, and a lot of it is clearly geared toward taking on human players. The AI doesn't really make you take advantage of a lot of the advice.

Thank you all for your help! I will start another game and see if I can get it to where I want it. I think I have AI aggressiveness at the lowest level, but I will restart and confirm. I want those silly invasions stopped.

[Gray Fox]

"Well, presumably you wouldn't have most of your CSA forces in West Virginia or deep in the Valley. "

Not so much...

viewtopic.php?t=34770&p=307797#p307797

[grimjaw]

jscott991, Athena is not exactly dumb, but the rules it plays by aren't something that would make much sense unless you could somehow see how they are laid out. At least that's what I hope. I only know what can be gleaned through observation, so I'm more of a naturalist than a biologist when it comes to this game.

The things that were supposed to help Athena were settings like detection and activation bonuses. But even if it can see what you're hiding in the woods and can act more often than you can, it still ends up doing things that seem to be autistic. I don't know what additional behaviors are enabled by that setting.

I'm only guessing here, but I suspect Athena makes decisions based on certain stats and not on personalities. I suspect it doesn't care who Stonewall is, but even if it did, it only has one playbook for a leader with Stonewall's stats. It has one or maybe two hammers (blue and grey ones, maybe?), but it's still just looking for nails. Using objectives you can give it more interesting places to hammer nails, but don't expect it to show up at those places with a power drill, a belt sander and some varnish.

Athena by itself I think doesn't make long term plans. It doesn't build a contingency force; it has to be scripted. There's so much than can happen on an open ended campaign that it would be difficult to plan for it all. Many games have this issue. (view this short example, about the one minute mark: https://www.youtube.com/watch?v=fBlGZJErn50 , or this one: https://www.youtube.com/watch?v=sFw6JW_A5XU ) CW2's full campaign has things scripted so that events from the real conflict happen on time, every time, even if they make no sense in the campaign you are playing.

Without changes to the rules Athena plays by, what's needed is a reassessment of its objectives periodically over the campaign. One or twice a year, an event would review things and say, hmm, based on this (wildly varying situation) our objectives should now be (completely open ended objective choices). It is possible to do this, within limits, but I am not at that point. Yet.

But the entity I am designing this script for is still going to be looking for nails to whack.

PBEM, though perhaps a frustrating method, is still going to be the most interesting way to play this game, or anything built on this engine.

[jscott991]

Athena doesn't have to be a genius to know to defend Richmond. I also don't get why it is obsessed with West Virginia, the Shenandoah, and western Maryland. So many CSA divisions just wasting away running around in those areas while I took Richmond with basically no losses.

I've played all the big Civil War grand strategy games (Grigsby's, FoF, this) and its remarkable how hard they all find it to simulate the basic course of the Civil War. Virtually every World War 2 game made gets the over all flow at least sort of right (invade Poland, invade France, invade the USSR). You would think a Civil War game would have it even easier, given the limited number of real theaters. Just make a Confederate AI that defends Richmond, Nashville, and Memphis and you're done. I don't get why so many CSA AIs goof around in WV, Maryland, or Missouri and lose their main objectives.

Grigsby's game is cartoonish, but at least it gets the main theaters right.

[jscott991]

The "epic" battle that ended my game.

Is this typical? Does the AI usually fail to defend Richmond?

[CWII_Monster]

Definitely can be. If you're looking for a challenge, let's do a multiplayer game.

[deguerra]

Hi jscott,

I'm glad you're still giving it a go - like I said, I'm well aware of its shortcomings but I still rate this game above all others for the Civil War - but not unconditionally.

Yes, Athena is (in my experience) not great at defending Richmond. She's not a bad AI, all things considered, but she does struggle with the magnitude of the task sometimes. However, there are some things you can do to help her:

- Game Settings: detailed more in this thread: http://www.ageod-forum.com/viewtopic.php?t=52916 but the main takeaway I've seen to be useful in my games is reducing the AI detection range to low and keeping the aggressiveness to normal at most. Beyond that, Athena seems to see threats everywhere and typically leads to the odd behaviour you're noticing of sending troups to far-off places.

- Some house rules for yourself: this is obviously highly personal, but I like to keep things vaguely historical anyway, and playing the Union at least this can tie in nicely with giving Athena a leg up. Keeping substantial garrisons in your rear, protecting Washington at all times and playing a little bit more like McClellan in the first couple of years generally means Athena's flaws are less exposed.

Overall I'd say Athena can be competent at executing a vaguely "historical" game, especially for a human Union player. However if you deviate from history by being more aggressive early on she will struggle. There are many fascinating "what-if" strategies to be explored in this game, but in my experience at least, those are best left to PBEM. Incidentally I can only encourage you to try a PBEM game. If only I had the time, I'd be playing far more of them as the people on this forum are fabulous

Hope this was of some help
-deguerra

[jscott991]

I actually played this game (and all Civil War games really) exactly like McClellan.

I kept a ton of troops in Washington until March of 1862, when I launched the Peninsula Campaign with three corps of three divisions each (ultimately three more divisions ended up down there for that battle above). I even left McDowell's big corps behind, just like in history. I slogged my way up the Peninsula as my generals ignored my order to advance, only making it to Richmond in late June of 1862. Then I finally got them to attack it in late July of 1862.

At no point did any Confederate troops arrive to help. As I showed in the screenshot earlier in the thread, they were running around in WV and the Shenandoah. From what I could tell, they had some troops that could have made this interesting, but none of them ever did anything.

I have aggressiveness set to zero (the absolute lowest). I've done this since CW1 when the AI Johnston used to run for Pittsburgh and just fall apart retreating back (depriving the CSA of half of its initial troops). I will try a low detection bonus next time.

What's funny is that the new Civil War game on Steam, Grand Tactician, has this exact same problem. The CSA rushes all of its troops to West Virginia, and you can take Richmond without firing a shot.

[AACW3Plz]

As stated, your cohesion problems are likely due to marching/weather penalties. Another issue might be troop mix - the Union spawns a ton of those 67 power volunteer units. They tend to lose cohesion quickly unless youre farming them under trainers (I have removed the ability from my game - I think it's grossly op) or mitigating penalties with militia/patriot leaders.

The AI usually responds to Richmond threats in my experience, but in fairness Ive only played a handful of Union campaigns. That said, yes the AI could be better. Given machine learning tech, hopefully strategy game AI will get a boost in the near future.

[pgr]

My pet theory is that the VP towns are drawing Athenia north. Specifically Winchester and HF. Oddly Fredericksburg and Manassas aren't considered strat towns, so the CSA does not mind abandoning them to make a play for WVa.

I finally figured out modding enough that I created a campaign with VP in Manassas and Fredericksburg, but not Winchester. I need to do a few campaigns to see if it helps the CSA stay centered in Va.

[jscott991]

My pet theory is that the VP towns are drawing Athenia north. Specifically Winchester and HF. Oddly Fredericksburg and Manassas aren't considered strat towns, so the CSA does not mind abandoning them to make a play for WVa.

I finally figured out modding enough that I created a campaign with VP in Manassas and Fredericksburg, but not Winchester. I need to do a few campaigns to see if it helps the CSA stay centered in Va.

Definitely let me know how successful this was. I don't know why it's so hard to keep the CSA defending its capital, but whatever can be modded to make them do it would be very helpful.

I started again and in October 1861, I'm already seeing the same issue. There's nothing in Manassas, Fredericksburg, and Richmond. And tons of AI containers in the Shenandoah and the wastes of WVA.

[grimjaw]

I'm willing to wager that you are never going to be satisfied with the single-player version of this game, for the same reasons I and any number of other people aren't. It's a long-standing issue that will likely never be addressed by the developers with this engine. See past complaints (and suggestions for trying PBEM). When even the majority of the community is uniting *against* the complaints about the AI ...

viewtopic.php?f=331&t=39488&p=346487

I am not willing to speculate about the reasons behind the lack of improvement other than to say it was probably related to business/economic considerations and not technical ones.

Having pursued computer science in education and later tinkering with it, I've always thought of programming as trying to teach the dumbest thing in the world how to add one and one. It has to be told

every
single
step
ad
infinitum

To create a system as open-ended and with as many options as a four year campaign like the US Civil War would be difficult enough. It has to be built with rules a computer can handle, rules even a human child can usually defeat, exploit or abuse. The computer also has to have the equivalent power, in today's relative terms, of a potato so as to cast the widest possible net for customers.

Take that Cyberpunk 2077 release that's in the news recently. It wasn't PC players that got it pulled from stores and refunds issued. PC players have never managed to do that even though that kind of thing has been happening for years now with games on PC. It was the cacophony of console players, which is where the market is. Probably things like facebook and Twitter contribute, too. PC gaming, PCs in general, started going the way of the dodo with the advent of smartphones (consoles never managed it). Shrinking market, fewer opportunities for making money, fewer reasons to expend resources, more of making due with less.

The same engine, fundamentally, running every AGE game; same AI fighting battles from antiquity to WW1 (thus my hammer and nails analogy). AGE isn't alone in this. I can think of any number of other games that have done the same thing. Many games were built on Valve's or id's engines, for example, and many sequels to games were retools of existing code.

With a more popular game there would probably be more modding to improve it. Skyrim's mod community is large enough that the ad revenue and voluntary contributions generated from website hosting the mods was enough to pay someone (or multiple someones) to write a program to install them. AGE doesn't even maintain its own wiki.

I've stuck with this game because the developers made the ability to mod it available, which I could not have done and am grateful for. I've only had time to fiddle with it when I'm not working. The limited improvements I have made or been involved with and uploaded to this forum, with the exception of the portraits mod, have generated almost zero feedback.

But why bother modding when I can get apps for free on my phone that I, like any other simian, can operate with my thumbs?

Sorry for the rant.

jm

[grimjaw]

I have been researching this anyway. In the course of things, I found some of the logic the AI is supposed to use to generate Richmond interest.

In the event file "1861 CSA AI.sct" there is the following event.

SelectFaction = CSA
SelectRegion = $Richmond_VA
StartEvent = evt_nam_CSA_LowerAggressivityMidAtlantic61_AI_ON|1|0|NULL|NULL|$Richmond_VA|NULL

Conditions

CheckAILevel = 1
MinDate = 1861/05/01
MaxDate = 1861/12/31

EvalRgnOwned = $Prince_George_MD;NOT

Actions
AI.SetAggro = $Mid_Atlantic;75;
NULL = NULL
AI.SetLocalInterest = $Prince_George_MD;50;
AI.SetLocalInterest = $Baltimore_MD;50;
AI.SetLocalInterest = $Wayne_Flat_WV;50;
AI.SetLocalInterest = $Westmoreland_PA;25;
AI.SetLocalInterest = $Philadelphia_PA;25;
NULL = NULL
AI.SetLocalInterest = $Fauquier_VA;200;
AI.SetLocalInterest = $Richmond_VA;150;

EndEvent

I'll assume some of you aren't familiar with the commands used. That event checks that three conditions are true:

1. CSA is being played by the AI.
2. The game date is between 05/01/1861-12/31/1861.
3. Washington DC region is not owned by the CSA.

Those aren't foolproof conditions, but they'll almost always work that early in the game.

If all three are true, the following things are set for the AI.

1. Stance (offensive-defensive) for the CSA is modified to 75 in the "Mid_Atlantic" area. According to the wiki page for that command, "A value > 100 is 'above normal', a value < 100 is 'below normal'". Mid_Atlantic is Virginia, but also New York, New Jersey, Pennsylvania, Delaware and Maryland. So any AI CSA forces in those areas should tend toward defensive postures.

2. A series of commands are given for several cities to set their interest for the AI. Among these, Richmond is only the second highest, after Manassas. I can see the logic behind some of the others, but I'm puzzled by Wayne Flat, WV; it's adjacent to Prestonburg, KY, and other than guarding that avenue to KY has nothing else of interest I'm aware of.

The following is from the wiki page of AI.SetLocalInterest:

"General Note: The value of a region depends of a dozen of factors, among them the structures types
and level, supply level, number of land links, Military control, part of a patch of controlled
regions (or isolated one), loyalty, Strategic City or Objective City tags and more...

The most important parameter: is the region a strategic city/objective?
A region of almost no interest can have a 2, while a coastal region with a mid-sized town and
harbor can get a 30 and the same but if defined as an objective can soar to 200!"

As defined by the April '61 scenario, Richmond is given an objective value of 10, twice as high as the next highest objectives. Richmond is also defined as a strategic city, which means it gives 1 VP per turn for holding it. As a result, Richmond, or rather the avenues by which Richmond is reached, should have a much higher value than other regions. The AI.SetLocalInterest command can be set as high as 999, so you could tinker with that value and see what it gets you. AI.SetLocalInterest is just for regions. There's a separate command for areas (groups of regions or other areas), AI.SetAreaLocalInterest, than can be used to set AI interest in such things as states.

A couple of things to note. I believe AI.SetLocalInterest and AI.SetAreaLocalInterest are set-and-forget commands, meaning they aren't per-turn settings. Once you run AI.SetAreaLocalInterest, it stays at the value you specified until the command is invoked again, or if there's some other command that operates on the same value. The event only runs once, probably on 05/01/1861, and those values will remain the same throughout the campaign. There's nothing else scripted to adapt to changes on the board. For the CSA, Washington DC is always "AI.SetLocalInterest = $Prince_George_MD;50;", and it's objective value is half that of Richmond (but higher than any other objective besides New Orleans and Charleston). The full campaign seems to want to steer the AI towards Richmond or DC, depending on which faction it's using, and is never given instructions to adapt after it achieves those objectives. The commands setting AI interest stop running at the end of '62, and even earlier for the Union. Promotion events for generals run to the end of the scenario, but that's it. Stance never changes in the mid-Atlantic area (AI.SetAggro = $Mid_Atlantic;75;). If Virginia is ever lost, the chances that are lower that the AI will seek to aggressively reclaim it, relative to other areas.

grimjaw's verdict: game designer believes single-player game is won in the first two years, after DC or Richmond are taken; doesn't expend effort coding for other eventualities. Full 1861-1866 single-player campaign is simultaneously WAD *and* delivered to customers incomplete. It's entirely possible that something was developed but not published with the game. As there are a number of unused events and scripts that *were* included, however, I think it is unlikely.

The original poster mentioned lack of forces defending Richmond. There is an event in the same file I mentioned above that spawns a large number of (permanently locked) units at Richmond, but only under certain conditions.

SelectFaction = CSA
StartEvent = evt_nam_CSA_DefendCapital_AI61-62|1|0|NULL|NULL|NULL|NULL

Conditions
MinDate = 1862/01/01
MaxDate = 1862/12/31

CheckAILevel = 1

SelectFaction = USA
SelectSubUnits = Area $Close_to_Richmond;FactionTags USA
EvalSubUnitCount = >=;70

As you can see, it only operates *after* 01/01/1862, and only for a calendar year. It's limited to detecting units in the area defined as "$Close_to_Richmond", which is the following regions:

$Spotsylvania_VA|$Caroline_VA|$King_and_Queen_VA|$Louisa_VA|$Buckingham_VA|$Amherst_VA|$Appomattox_VA|$Henrico_VA|$Charles_City_VA|$New_Kent_VA|$Williamsburg_VA|$Prince_George_VA|$Richmond_VA|$Surry_VA|$Waverly_VA|$Dinwiddie_VA|$Camden_VA|

Roughly, that's the regions immediately adjacent to Richmond amd from Fredericksburg down to the peninsula. Not Norfolk or Suffolk, though, leaving a backdoor corridor around the AI's response if you want to cut it off via North Carolina. If you want to take Richmond, you can game the system by doing it before or after 1862. The number of elements has to be equal to or greater than 70. If you launched a deep cavalry raid, up to three full AGE divisions (54 sub-units, if you include the leaders) wouldn't trigger the event.

The AI was never scripted to adapt past the end of '62, and the game in general was not scripted very well to adapt to changes on the board. In the game I am testing now as the CSA, I've held Alexandria, VA, with a large force for two months, but the event that spawns many Union brigades still loads them (locked) in Alexandria. 19th century paratroopers, who'd have thunk it?

Although the game does include emergency militia events to defend against incursions into states, it doesn't have any scripting to adapt to a moving front line. The CSA objectives and their weights are:

$Richmond_VA;10;$Prince_George_MD;5;$Shelby_TN;2;$Iberville_LA;5;$Davidson_TN;3;$Atlanta_GA;3;$Saint_Louis_MO;2;$Arkansas_AR;1;$Mobile_AL;3;$Norfolk_VA;1;$Warren_MS;1;$Baltimore_MD;1;$El_Paso_TX;1;$Charleston_SC;5;$Jefferson_KY;1;$Chatham_GA;2;$Bexar_TX;1

Nothing about that ever changes. The AI might have lost Arkansas and Mississippi, but St. Louis will always be a higher objective. I am trying to improve upon this but I have a dog and a job, neither of which pays me very much. Especially the dog.

[AACW3Plz]

I am probably not wealthy enough to pay you exclusively for your time (turns out kids are expensive), but I'd certainly be willing to chip in on some sort of AI overhaul project if there was a kickstarter or patreon. We may not get an AACW3, but a 2.5 might suffice.

This game really could use a refresh. I've been doing minor things like tweaking the Generals and trying to simulate the campaign start more effectively (probably going to get rid of 90% of the regulars and possibly even 3 stars until post Manassas to simulate that shitshow, why the CSA didnt just march on DC, and the import of Mac's promotion), but like you I'm pretty busy and tbh Im a bit out of my depth on some of this.

[Blood and Thunder Brigade]

I am probably not wealthy enough to pay you exclusively for your time (turns out kids are expensive), but I'd certainly be willing to chip in on some sort of AI overhaul project if there was a kickstarter or patreon. We may not get an AACW3, but a 2.5 might suffice.

This game really could use a refresh. I've been doing minor things like tweaking the Generals and trying to simulate the campaign start more effectively (probably going to get rid of 90% of the regulars and possibly even 3 stars until post Manassas to simulate that shitshow, why the CSA didnt just march on DC, and the import of Mac's promotion), but like you I'm pretty busy and tbh Im a bit out of my depth on some of this.

Patreon might be one way of doing it, yeah. I'd certainly contribute to such a thing and I'm sure others would, too.
As for a refresh, perhaps with enough 'hands on deck' then problems that might be a bit too much for an individual might not be a problem for a collective. A 'many hands make light work' approach so to speak. And if a CW 2.5 was sufficiently impressive then perhaps a CW3 might suddenly be a real possibility.

[Majorc28]

I would support this endeavor with some cash or time. Whatever is needed. I am not a programmer but I can test mods, research unit names, etc. What about setting up a Discord server? This forum style is very dated and Discord is a much better way for rapid communication.

[Aerosol]

I am probably not wealthy enough to pay you exclusively for your time (turns out kids are expensive), but I'd certainly be willing to chip in on some sort of AI overhaul project if there was a kickstarter or patreon. We may not get an AACW3, but a 2.5 might suffice.

This game really could use a refresh. I've been doing minor things like tweaking the Generals and trying to simulate the campaign start more effectively (probably going to get rid of 90% of the regulars and possibly even 3 stars until post Manassas to simulate that shitshow, why the CSA didnt just march on DC, and the import of Mac's promotion), but like you I'm pretty busy and tbh Im a bit out of my depth on some of this.

I second this and would also be willing to contribute to an AI overhaul project.

If you were somehow able to create a France v. Mexico sub-campaign well then you will have set my posture to 'aggressive'..

[grimjaw]

Before you go down the concerted effort track, I'd suggest that you come to a consensus, and decide if all that's required to achieve the end you want is skull sweat and labor.

I do not believe the owner of the engine code, whoever that may be, is going to open it for free to all and sundry. Even if they did, they're unlikely to officially approve mods I produce because besides model and units files I don't and won't use spreadsheets to update the scripts. I'm directly editing with a decent text editor I've customized to highlight keywords. I've kept previous revisions of scripts and I'm trying to document as I go along but I'm not using a formal system.

I can create using their scripting system, but I don't know what's under the hood in the engine, doubt i personally could do much to improve that in a reasonable timeframe. I will say though that the scripting system is very limited in some ways. AFAIK, there's no provision for evaluating conditional statements or perform actions using OR. It's harder to get the AI to react like you want via script as a result. It leads to a bulky mess and I suspect alot of churning by the engine. I'm trying to see if consolidating redundant actions will help speed up script processing. For example, the script to unblock Kentucky uses the same block of text five times, which I'm testing to reduce to one. If that works, the same can probably be done for other events. I don't know yet if a script file can be called by actions in an event. If they can it might be possible to simplify editing, and maybe reduce script processing time. That last bit would just be icing. I don't really feel hampered by turn processing times currently.

It is a lot easier to improve the campaigns for PBEM because you can rely on humans to work out agreements in places where the game breaks down. For instance, to simulate lack of divisions early in the conflict players can agree to form no divisions over X elements. But to do any better for that for the AI requires brain surgery.

I am trying to improve the game via scripting, and I'm trying to bring the AI along for the ride. This involves compromises that I don't have to account for with humans. It's possible to create a campaign tailored to human players that dispenses with many events that exist mainly to keep humans from walking all over the AI. The developers included steps to cull those events in PBEM games, but the scripts still contain all of it.

I believe they stopped short of trying to improve the AI's response. Why? It is already remarkably hard to get a computer to do what we're talking about, so it might have been that. Too much time involved because they'd already made bank off the franchise and they'd rather be in the south of France enjoying the weather? Also possible. Maybe both, maybe neither. But helpful to know before much time, treasure and blood is expended. I personally fiddle with this thing because I perversely enjoy tinkering for it's own sake.

Right now I am proceeding along a path to improve my own personal game experience, but I am trying to do this in a way that lets me back up. Rather than rewrite events and scripts I'm writing new ones. I don't have to please anyone else. Anything other than that requires consensus. The common consensus I've seen on this forum, especially from beta testers or people who had direct contact with the developers, was 'if it's broken it ain't going to be fixed because that would require too much work.' In this thread there are people offering to pay for work that the original crew weren't motivated to perform. Think about that before you reach for your wallet.

All that aside, after improvements this is still going to be an AGE game, which means it's a game of regions and strategy and "NOT* batttlefield tactics. As a result I think it's a poor choice for US Civil War "simulation* because much of what is interesting about that conflict is precisely what the game can't deliver: tactical combat. A game where flanks don't exist can hardly expect to reproduce Chancellorville.

[pgr]

I'd suggest people shifting this conversation to the mods sub forum if we really want to get into the nuts and bolts of what's possible.

[jscott991]

I've now played four games over the holiday through about mid-1862.

In every single one, the AI will NOT defend Richmond. It is absolutely obsessed with West Virginia and the Shenandoah, but it will not stop me from advancing up the Peninsula and taking Richmond without a fight. And it rarely interposes itself between Washington and Richmond to stop an overland advance by other units.

There are some things going on that I don't entirely understand. (Just as a note, I play with FoW off so I can watch what the AI is doing.)

As I was advancing up the Peninsula in a few games, Beauregard or Johnston's army would appear in Richmond on one turn, only to leave the next turn. It never remained in position to fight McClellan's corps when I finally made it to Richmond.

From time to time, a stack of units called Robert E. Lee's command would appear in Richmond, giving the appearance it was defended. But that stack never stayed to stop me from taking Richmond without a fight.

I believe Richmond's garrison is activated in some way so it can move, and then the CSA AI moves it out to do other things in western Virginia.

I'm sorry to be repetitive, but I just have some questions.

1. What draws the AI to western Virginia? Why does it defend that territory more consistently than Richmond? Is there any way to easily change that AI priority so all those troops stick around Richmond or in the region between Richmond and Washington, D.C.?

2. How are other people playing this game at all if the AI simply concedes in the main theater? I got so many recommendations to try this game, buy it, give it a chance, etc. both in the last month or so and back when I first bought it, that I can't understand why it is so broken. I can even find old posts of mine about the CSA's Pittsburgh obsession in CW1 where people say that the AI doesn't do that kind of thing any more. It most certainly does!

[Majorc28]

I think the fog of war being off might be screwing up the AI's detection. Try playing with it on and see if you get some better results. The AI will usually defend Richmond somewhat well in most of my games. At least not just let me march in without a battle or two.

[grimjaw]

As I'm sure you know, there was some thrashing around in western Virginia by rebel forces in '61. Anywhere from Phillipi in June to Camp Allegheny in mid-December. That the AI would expend some effort there wouldn't bother me, but I understand your frustration that it doesn't seem to be focusing on more important objectives.

I don't know how the AI deals with FoW, actually. Maybe somebody reading this can give a definitive answer. The toggle switch for FoW in the game says, "Fog of War will prevent undetected units from being seen by the enemy" but that's obviously not the entire effect, or maybe the actual effect. For the player it means, among other things, that I can see power ratings of any unit on the board. But it won't tell me supply levels in regions I don't control.

I believe you can enable FoW for yourself while disabling it for the AI by exploiting the fact that you have the control over the toggle switch and it doesn't. Before the AI has generated orders on a turn, disable FoW. Issue all of your orders, enable FoW, then process the turn. If your order files didn't include a flag that says "all these orders were issued with FoW turned off", (and if that flag overrides the toggle switch it would be an 'undocumented feature') then the AI is going to process the turn with whatever setting is running at the time it generates orders.

Coincidentally, unless there's a flag in the turn files that identifies what FoW setting was in effect at the time orders were issued, you don't know if a human player has cheated because they can always flip that switch without you knowing. Not that it matters, because the turn files are plain text and could always be edited to change the flag.

In my own games I've been playing with FoW on, and I haven't seen the exact behavior you described. The Richmond, VA region has some forces defending it in my current playthrough, but the area through which the Union would invade is not heavily defended. Magruder's forces went to Richmond instead of the peninsula, and Huger is sitting in Suffolk, content to count the ships in Norfolk via a set of binoculars.

If the AI doesn't know what you're going to do or where you are, what strategy, specifically, should it be following? Given those conditions, should the AI follow the same strategy every time? Should it have a set of strategies to choose from, randomly or based on a weighted system? If Joe Player knows those strategies, how is the AI going to adapt to what amounts to Lee's Lost Order? It's even worse for the the game robot Robert E. Lee because Joe Player doesn't have to rely on cavalry, semaphore and the telegraph to coordinate his forces.

Think about how much planning we're talking about and the level of contingencies involved in trying to get a computer to respond intelligently to input it hasn't been programmed for, and keep that in mind the next time you see a driver-less automobile.

What you need to know are the rules that the AI uses to issue orders in absence of other guidance from scripting. The engine is probably coded to handle a set of generic situations, because it's been ported to several different scenarios. Since I've already pointed out that the AI in CW2 was not scripted for defense in depth, or for dates past 1862, I believe it's following the rules for generic situations. One of the developers, or someone who *knows* how this works under the hood, would have to chime in at this point to tell us what the rules are. Otherwise I don't think you're going to get a satisfactory answer. I've only seen vague references to these rules, never seen them laid out, but I am *very* curious to know what they are.

Speaking of objectives, you could always be mean to the engine by tweaking the full campaign scenario to remove *all* objective values. That's as mean as cutting off a cat's whiskers, but you might get an idea of what it wants to do generally.

[jscott991]

Your posts are informative, but you talk about strategy game AI as though it's a Chimera that can never exist (perhaps, more accurately, a unicorn). But as I said earlier in the thread, has anyone ever played a WWII strategy game where the early part of the war, at least, doesn't at least superficially resemble the actual war (meaning an invasion of Poland, a battle of France, usually a war with the USSR)? WWII games are just an example. I've played tons of wargames in my time on the PC, and few have been as completely unsatisfying as this one. Heck, just where Civil War games are concerned Grigsby's game and FoF do a much better job (the latter much worse than the former) of putting up some semblance of a Confederate defense.

I don't see why it's an impossible dream to have an AI that defends Richmond. Let its army sit in Manassas until it "detects" a Union force on the Peninsula (or wheverer) that is threatening Richmond. Then let it zip down and defend it. How hard is that? The CSA's armies are already flying around the map randomly (presumably using rail movement), shifting between the Shenandoah, west Virginia, and, sometimes, Manassas.

Just freaking defend Richmond. It's not that hard.

As for the west Virginia issue, Confederate forces in this area never exceeded much more than 5,000 men, and McClellan (later Rosecrans) conquered it with ease. I consistently see stacks with strengths in excess of 1,000 (usually Johnston and Beauregard's entire armies, supplemented in 1862 with an occasional corps) sitting in west Virginia. This is just absurd. There's no historic or logical reason for any of them to be there.

As it is, there is no eastern theater to speak of. I take Richmond around July of 1862 every time. The one time I played much beyond that, I was able to penetrate deep into North Carolina and link up with Washington by land . . . all without fighting a single real battle.