AGEOD Forums

I beat the Rebs by Nov '62. Default settings. Won some big battles in the Va countryside as Johnston kept flinging his army into my well-entrenched defenses. Things worked well. Gotta tweak the settings I guess, because I'm really not very good (my army's a mess).

What I didn't expect was the win was from high Northern morale, not low southern morale. Lots of ways to win or lose this game.

If you play as Union, I would say you should use the Hard level AI with one point of FOW setting, to even a bit things. And wait for the 1.06 AI, she will be better.

Pocus wrote:If you play as Union, I would say you should use the Hard level AI with one point of FOW setting, to even a bit things. And wait for the 1.06 AI, she will be better.

I'm trying one as the CSA now. Much harder for obvious reasons. I'll see what happens with that, then tune it up. I notched down the agressiveness a bit. Is that a bad thing to do? I want the Union to kick my butt, but I don't want it to feel like they're cheating, if you get my meaning.

I beat the Yankees by March '62. Default settings. [...]

Link

Asa wrote:I beat the Yankees by March '62. Default settings. [...]

Link

Nice. I guess you went straight for Washington?

Real wargamers play humans.

Crimguy wrote:Nice. I guess you went straight for Washington?

The Yankees went straight for Richmond! They were surrounded and destroyed (giving me 142 for NM by the end of November 61)... Nothing stood in the way to Washington.


Runyan99

those units "destroyed" - the icons in the casualty battle report - are those regiments or whole brigades ?

runyan99 wrote:Real wargamers play humans.

Real wargames can be played solitaire.

gwgardner wrote:Real wargames can be played solitaire.

I disagree. There is no such thing as a smart AI.

Only dumb players.

Hmm, I think it's safe to say I have no chance of ever defeating the AI in Fritz and Chessmaster 10. In fact, I'd wager none of us in this forum will ever do so.

And it's only a matter of time before wargame AI becomes as unbeatable as the chess AIs. I think we've already seen some significant strides in AI in games in the past few years: Galactic Civilizations II has a pretty impressive AI, Conquest of the Aegean has a decent AI, and AACW itself has a solid AI. The market supports this: most people play the AI, not PBEM.

Besides, often we play an AI so that we can replay the war in a plausible way, not just so that we can thrash an easy opponent. I enjoy PBEM, but my AI games actually play out more historically. When I play War in the Pacific against the AI, I impose house rules on myself and play in a "historical" fashion, which is what the AI expects. It can result in a very interesting and entertaining game.

Moreover, there's a long tradition of playing serious wargames solitaire. I played Advanced Squad Leader (and Solitaire ASL) solo for years. Now, with the PC, we can play solitaire and, if need be, give the AI some advantages to increase the challenge. Or we can switch sides after a few turns.

I've played plenty of PBEM (Uncommon Valor, WITP) and face-to-face boardgames (ASL, Third Reich, many other Avalon Hill games). Human opponents are great. But there's room in gaming for both human and computer opponents.

I´ve been playing the Union with 1.06, there´s a notable improvment of the AI but still got a morale victory by mid 1862 when Richmond fell.

As far as I know this is plausible scenario from the historical point of view, but there should be an option to alter the victory conditions without resorting to give the AI same bonuses to make the game harder (if there is one I haven`t found it), ot maybe a "what if" scenario where the UK and/or France intervene in favor of the Confederates from the start.

One problem is the casualties rate, it is terrible, the AI keeps committing suicide sometimes, is like 3 to 1 in my favor, this must be quite unhistorical, but i´m not an expert.

On the other hand the AI seems to build better armies and concentrate them in more logical ways for the most part.

pablius wrote:I´ve playing the Union with 1.06, there´s a notable improvment of the AI but still got a morale victory by mid 1862 when Richmond fell.

As far as I know this is plausible scenario from the historical point of view, but there should be an option to alter the victory conditions without resorting to give the AI same bonuses to make the game harder (if there is one I haven`t found it), ot maybe a "what if" scenario where the UK and/or France intervene in favor of the Confederates from the start.

One problem is the casualties rate, it is terrible, the AI keeps committing suicide sometimes, is like 3 to 1 in my favor, this must be quite unhistorical, but i´m not an expert.

On the other hand the AI seems to build better armies and concentrate them in more logical ways for the most part.

The 1.06 readme says the AI is yet to be fully optimized for defense of capital. I hope it will be done in a few days...

Let poor Pocus at least get back from his vacation, eh?

Yes, I'm wondering if 'low' aggression might be the key at least until the capital changes and maybe even after then.

I also beat the ai in my first game, playing the CSA in the April 1861 campaign and taking Washington, DC in November '61. That was on normal/low detect bonus (1.05f). I then tried as the USA and won by mid '62 - just wearing the CSA down with lost battles everywhere. That was on normal/very high detect bonus (1.06). I switched back to the rebs and won again by November '61, taking Washington on Hard/Very High. The trouble is McDowell wanders off to the valley, leaving Scott behind and not enough defenses, especially with the command penalties they suffer. They're better at building stacks but still don't have the perfect no command penalties anywhere that a player can.

I decided as it's a November victory that I can test this as the CSA easier so then tried Hard/Very High again but with 'low' aggression. It was very encouraging - I only played the first 10-12 turns or so before I invaded Kentucky by accident - fighting in Columbus, which should be my area but oh well. McDowell has remained in Washington, the ai has collected the US cavalry that starts out on the west and raids Missouri/Arkansas/Texas in to a couple of stacks instead of each staying individually so is much harder to beat. No invasion of Kentucky, which the ai does (as either side) in June 1861 on 'normal'.

It's encouraged me that this may be the better setting - I'm probably going to try again and actually lower the 'detect' bonus down a little as I'm wondering if this is making the ai flood the atlantic blockade box, killing anything I send there immediately but leaving the gulf blockade alone.

I especially think it will be better when the AI is CSA - stop the Fort Pickens disastrous assaults, early invasion of Kentucky and general suicide attacks (which is much more important with their manpower). I'm not so sure about when the AI is USA for the long game but hopefully I'll find out and it won't be a foregone conclusion come the end of '61

Grotius wrote:Hmm, I think it's safe to say I have no chance of ever defeating the AI in Fritz and Chessmaster 10. In fact, I'd wager none of us in this forum will ever do so.

And it's only a matter of time before wargame AI becomes as unbeatable as the chess AIs. I think we've already seen some significant strides in AI in games in the past few years: Galactic Civilizations II has a pretty impressive AI, Conquest of the Aegean has a decent AI, and AACW itself has a solid AI. The market supports this: most people play the AI, not PBEM.

Besides, often we play an AI so that we can replay the war in a plausible way, not just so that we can thrash an easy opponent. I enjoy PBEM, but my AI games actually play out more historically. When I play War in the Pacific against the AI, I impose house rules on myself and play in a "historical" fashion, which is what the AI expects. It can result in a very interesting and entertaining game.

Moreover, there's a long tradition of playing serious wargames solitaire. I played Advanced Squad Leader (and Solitaire ASL) solo for years. Now, with the PC, we can play solitaire and, if need be, give the AI some advantages to increase the challenge. Or we can switch sides after a few turns.

I've played plenty of PBEM (Uncommon Valor, WITP) and face-to-face boardgames (ASL, Third Reich, many other Avalon Hill games). Human opponents are great. But there's room in gaming for both human and computer opponents.

I think COTA has the best ai in wargame currently. I always wondered if Galactic Civ II was cheating - it always seemed to know where the best planets were. WitP doesn't know what to do if you try something novel - I've had to impose the same house rules as you.

And btw, I've tied Fritz a couple times, never beat it without turning down it's big brain. Out of possibly 5000+ games against it. I think that's luck rearing its ugly head. With proper endgame hashtables loaded up I would have lost I think.

Wow, you got a draw against Fritz with its brainpower fully unleashed? I have no hope against it, but then I'm a mediocre chess player -- rating 1500 or so.

Until Pocus optimizes the capitol-defense AI, in my next game (as the CSA), I'm going to impose a house rule on myself: no early capture of Washington, DC. I'm really less interested in winning than in playing a historical re-enactment of the war, if you will.

Change the names on the AI settings...Cautious becomes "experienced" , "normal" becomes "novice" and "aggressive" becomes "reckless".

Grotius wrote:Wow, you got a draw against Fritz with its brainpower fully unleashed? I have no hope against it, but then I'm a mediocre chess player -- rating 1500 or so.

Until Pocus optimizes the capitol-defense AI, in my next game (as the CSA), I'm going to impose a house rule on myself: no early capture of Washington, DC. I'm really less interested in winning than in playing a historical re-enactment of the war, if you will.

You're about at my level. I'm a 1500-1600 at my best, and usually much worse (I'm a very lazy player). You're probably better since I used to play a lot against Fritz and still only managed these 2-3 draws.

Well, the hard/medium detect setting with low aggression game I played didn't work out in the end. The ai lost a large part of the Army of the Potomac in Stafford, VA (others have mentioned in the 'greatest victory' thread how it does this). Given that, Washington fell in December 1861 and the game was over.

Still, that's a month later than normal I'm generally happier with the amount being kept near Washington at 'low' though so it's back to hard/very high and if the same happens then I'll play the Union until the next 'capital protection' improvement

I played as Union in 1.05a demo, normal settings. I played very cautiously avoiding any major battles. It was a cat and mouse game with both forces retreating a lot. I was able to take Manassas, Stafford region in VA and Fredericksburg in January 1862 while fighting only one big battle (I lost 5000 soldiers to CSA's 10 000).

Two observations:

1. CSA assaulted many times Ft Monroe and Pickens with heavy losses. This behaviour was mentioned many times in other threads. AI should lay sieges and built more arty instead. At Ft Monroe it attacked with 6 elements of cavalry. At Ft Pickens CSA had 20 elements under gen. Charles Winder againts my 3 elements, so at least there it built a considerable force. Still, it should just lay siege or
check my garrison with some militia located nearby. Almost 30-40% of all CSA total losess in the game could be attributed to those stupid assaults!

2. Virginia. The CSA AI made the same mistake as I have while playing the Confederates: paying too much attention to holding Harper's Ferry and Winchester. They are both strategic cities but holding them is COMPLETELY useless. The whole Eastern theatre lies on Richmond - Philadelphia axis. PERIOD. The CSA AI was cut off from supplies by first cutting their railroads in Manassas and Stafford, VA and later by occupying those regions with strong Union stacks. From there you can quickly take Charlottesville, Fredericksburg and Richmond, while Beauregard, Jackson and Johnston look for bread in Harper's Ferry and Winchester.

On a positive note, CSA moved its capital to New Orleans early in the game and guarded it with a strong brigade force lead by general Sibley.


After the game ended, I switched sides to examine CSA predicament. Well, its railroad/river/naval capacity was at almost 0. It had almost 300 k dollars ,772 (sic!) companies, and 50 WS. If AI neglected transport, it could easily recruit those massive VA brigades instead to halt my approach, but it didn't.
It seems the AI is completely unable of guessing how much recruitment is going on on the other side and reacts only to actions made by the player, while it should be suspecting that the player is bulidng massive armies beyond the frontline.

The AI got a major boost with the 1.06 patch, so it becomes a bit moot to evaluate its performance with the 1.05-patches

Jacek wrote:1. CSA assaulted many times Ft Monroe and Pickens with heavy losses. This behaviour was mentioned many times in other threads. AI should lay sieges and built more arty instead. At Ft Monroe it attacked with 6 units of cavalry. At Ft Pickens CSA had 20 units under gen. Charles Winder againts my 3 units, so at least there it built a considerable force. Still, it should just lay siege or
check my garrison with some militia located nearby. Almost 30-40% of all CSA total losess in the game could be attributed to those stupid assaults!

The Union garrisons in these locations are actually 'outside' the forts - odd I know but actually better because they'll fight outside with a heavy bonus and be able to retreat inside if they lose. It means anyone moving on to the forts can't just seige but has to fight first so the AI can't just siege. It should just leave these locations alone though.

Jacek wrote:2. Virginia. The CSA AI made the same mistake as I have while playing the Confederates: paying too much attention to holding Harper's Ferry and Winchester. They are both strategic cities but holding them is COMPLETELY useless. The whole Eastern theatre lies on Richmond - Philadelphia axis. PERIOD. The CSA AI was cut off from supplies by first cutting their railroads in Manassas and Stafford, VA and later by occupying those regions with strong Union stacks. From there you can quickly take Charlottesville, Fredericksburg and Richmond, while Beauregard, Jackson and Johnston look for bread in Harper's Ferry and Winchester.

.

I agree with this observation. I have seen the same behavior many times. The weighting of Harpers Ferry and Winchester as strategic cities is evidently too high, which causes Confederate AI to value them too much. They deploy the armies there, rather than Fredericksburg, leaving the road to Richmond open to the Union.

IMHO Fredericksburg should be greatly increased in its VP/strategic city value to help the AI. This would also be historic as well. Keeping the Federals on the other side of the Rappahannock was a huge goal for the Rebels, who fought several battles there to do just that, including Fredericksburg, Chancellorsville, the Wilderness, and Brandy Station. The strategic goal of Lee in all of those battles was to throw the Union back across the other side of the Rappahannock. Lee knew once the Union was able to cross the Rappahannock in force and stay on the other side, it was only a matter of time before Richmond fell. Grant proved this in 1864, making one of the most critical decisions of the war to keep pushing forward after initial tactical defeat in the Wilderness, producing the strategic victory of the Forty Days.

I would respectfully petition the designers to consider making these changes of devaluing Winchester and Harpers Ferry and increasing the value of Fredericksburg greatly.

Hancock, I am with you in petitiong the devs 'bout Harper's Ferry and Winchester. I think with this change even on Normal setting you could experience a greater challenge as the Union.

Do we have the ability to mod VP sites? If so, would be good to test this out before asking for it to be official.

Sure; I saw it described in a thread somewhere around here

EDIT: Here it is: http://www.ageod-forum.com/showpost.php?p=39332&postcount=8

I was the one asking about that before. I've experimented with some limited success, and a few colossal failures. Before getting too hard core into it I'm going to await the capital defense update. I think that combined with reducing Harper's Ferry (which is in line historically, only Winfield Scott thought it critical early in the war...oh read Dissonance for a great discourse on that). Also increase Fred. will also help keep rebel armies in a better position. Finally I'm going to hopefully address my other issues with Memphis and Nashville....reducing them. Then increasing Vicksburg and Atlanta...maybe Baton Rougue and Charleston too. My goal is to get the critical cities away from the borders, the Harpers Ferry and Fred. ideas would really add to this.

I think VP weighting might help. Not sure how much of the answer it is, though. The reason I say this is because a CSA player will go straight at Washington, knowing the Federal AI won't be aggressive and head south. Thus he won't care too much about the VP value of Fredericksburg or Richmond, unless the Union AI is made to respond to such VP values. Maybe it could be - that would be great. But I'd also like to not see a Federal AI that gets too crazy early on and loses its numerical superiority, which is the only buffer between the Rebs and DC.

To extend the game as the CSA I came up with a house rule, I´m not going after Washington, but I´ll encircle it...So far it is working, the game is still going by winter 63/64 with the federal capital completly sorrounded by Lee´s Army.

The Union army, cut off from Washington, tries again and again to brake the the lines in huge battles in the Annapolis area, but so far my coprs have held the line. The campaign in the west has been fun, with my armies now marching on in Sprinfield and Indianapolis while keeping in check a couple of big Union´s commands.

The downside is that while the AI produces HQs in Washington it doesn´t send them to the west, maybe because it perceives the threat and keeps them in DC as part of the defense effort, wich is wrong since they would be much more usefull in the west.

And for whatever reason a bunch of Union generals and 40 thousand troops are in one of the forts south of Florida doing nothing.