Idea: Since the campaign AI is moddable...

[Parallel Pain]

Why not make it competent?

Right now it does this:

Step 1) Make an army big enough and high quality enough to match the enemy force in province X, a border province of an enemy
Step 2) Move into province X and attack any enemy army in it
Step 3) After all the field army in that province capable of marching to save the settlement is gone, lay siege to the settlement
Step 4) Take the settlement
Note: If the army loose a battle, move back to friendly province and get more troops.
And also if any enemy army is in own territory, send all available force after it.

Right now that gets exploited to death, with players (at least me) occupying all the bridges, and move a good stack into the enemy territory and occupy the high ground and destroys everything that comes at them.

So making the Seleucids as an example:
Step 1) It decides to focus on the Ptolemy
Step 2) It builds up offensive armies on provinces bordering the Ptolemy.
Step 3) It sends token stacks (enough for defense, not enough for offense) to occupy river crossings and build forts on high ground for a defensive stance against all other enemy faction.
Step 4) It sends it's large stacks into the Ptolemy territory to occupy any river crossing and high ground to try to force the Ptolemy into a decisive battle.
Step 5) It finally moves to siege the city
Note: If it is defeated it tries to move toward the nearest friendly army OR the nearest friendly province. If it meets the nearest friendly army the two either join or move together.

If the Ptolemy move its stacks into a province and occupies bridge and highground. Send a stack and plant it near the Ptolemy stack to keep an eye on it, but not to attack until a sufficient force has been gathered to win (or so it thinks). The Ptolemy will also be programmed to do this, and so any Seleucid stacks in Ptolemy province would also have to judge if it can beat its shadowing opponent on the ground its standing on. If it can, attack. If it can't well, don't wait to be annihilated unless there's no choice, move to some even better ground or co-ordinate with other stacks for gang-up or bypass them and head toward the settlement and lay siege to it to force a battle on chosen ground.


And one easy way not to break an alliance is not to target any province of an ally border or not unless there are no other bordering factions that's not an ally, in which case choose one and not all at once. And don't choose the player's province at the first available opportunity.

Right now it's a grind game with the AI just throwing army after army directly at the problem with no brains what so ever. I don't want a grind game. After playing half a campaign it really tests your patience (at least mine) as you do the same thing over and over and over again.

Oh and of course throw out the "I will fight to the death always, you haven't beaten me even if I only have one settlement with 50men (not units, men) left.

If I was the Seleucid, and I destroyed all the Ptolemy stacks laying between me and Alexandria, I expect the Ptolemy to be begging for mercy and willing to pay tribute and all that. After all it's a lot better than loosing Alexandria. And it won't attack me again until it has rebuilt sufficient strength.


Hopefully with that it'll turn into a game that challenges your brain instead of one that tests your patience (Don't want EB to turn into an MMORPG experience after all).


Now if only the battlefield AI is moddable too. Sick of doing the same thing over and over and over and over again on the battlefield as well.

[Redmeth]

These are good ideas, but we don't have anyone specialized on AI modding to implement them right now. Quite a few months ago we tried contacting a few people who were investigating the AI but it didn't pan out due to their lack of time IIRC.
So, if anyone reads this and feels like they have a grip on AI modding on M2TW and has the time to experiment and implement it we do need the help.

[Lusted]

Battle AI is moddable as well, though limited, just like with the campaign AI. You can make the AI want peace more readily, but you cannot make it give tribute etc., and there is noly so much that can be modded. Check out these 2 threads for discussions on campaign AI and modding it:

http://rtw.heavengames.com/cgi-bin/forums/display.cgi?action=ct&f=9,6718,,10
https://forums.totalwar.org/vb/showthread.php?t=95416

And this thread has good tips for general AI stuff:

https://forums.totalwar.org/vb/showthread.php?t=83709

[The General]

... More bridge battles? :no:

[Parallel Pain]

That's good for a challenge. Or you'll have to use some brains to bypass the bridge.

I still remember shogun where you HAVE to fight bridges, that was FUN.

When there's only TWO bridge battle in the whole of a 150 year campaign in which I had to attack, and during one battle instead of blocking the bridge the AI decided to occupy the high ground, and during the other not only did it not block the secondary ford, after I crossed it still didn't send troops to meet me and just stood there as I moved my HA behind it and shot it to pieces and it still just stood there...

It tells me a lot about the competence of AI.

[woad&fangs]

IIRC the AI does not know about the existance of the second ford.

[Cartaphilus]

... More bridge battles? :no:

No, please!!!!!!

:no:

I've fought hundreds of ford/bridge battles in EB, and I am tired of them.
So in the future I'll do as the saxons in Maldon.

[Parallel Pain]

Use your brains, it's a strategy game.

And yes woad&fang, just while that explains why they didn't guard it, it doesn't explain why they don't send troops to meet me once I'm on the other side (and on their rear too).

Besides, while having defended crossings, I've never seen the AI smart enough to permanently guard a strategic crossing with a sizable force, and certainly when they do accidentally put a stack on it, just moving beside the stack and ending turn will make it either move off or attack you. And there are plenty of places to bypass crossings.

And heck it's nice if you have to assault once in a while for a change.

[Lysimachos]

IIRC the AI does not know about the existance of the second ford.

I can only speak for RTW on this, but it does know about the existance of the second ford. I have seen the AI use the ford, but only when i had positioned troops to guard it...

[Cartaphilus]

I can only speak for RTW on this, but it does know about the existance of the second ford. I have seen the AI use the ford, but only when i had positioned troops to guard it...

And not always.

Usually they don't do anything with the second ford, although you guard it. Only AI tries to force it, when AI has two armies and the second one (the reinforcements) passes near the ford.

[Elmetiacos]

The problem with bridge battles at the moment is that the AI always seems to commit his general to fighting on the bridge, which almost inevitably gets him killed.

[Jolt]

The problem with bridge battles at the moment is that the AI always seems to commit his general to fighting on the bridge, which almost inevitably gets him killed.

Sadly true. The computer often sends the General unit right into the midst of a overly compact fight (Such as bridges and fords) and then the A.I. must be indiferent to the casualties that the General unit is suffering. While any decent player who sends his General into such a tight battle, the player normally tries to withdraw the unit once the casualties begin mounting up or increase rapidly. The A.I. just let's the General fight to his death. I had some very funny times because of it, as the entire opposing army fighting in the ford (Full stack vs Full stack) routed 5 seconds after the General died. I wish I took pictures of the river in that battle.

[Elmetiacos]

I don't have MTW2, I'm not even sure I've got enough disk space, but I'd be interested in looking at the file(s) for the AI, because I've had some success modding AI for files for Europa Universalis and Age of Empires. Is there any way I could look at MTW2's AI and make suggestions?

[Lusted]

This thread gives an overview of AI modding:

http://rtw.heavengames.com/cgi-bin/forums/display.cgi?action=ct&f=9,6718,,10

[Cambyses]

Well, historically, a great many battles were fought at river crossings. In fact it would be interesting to get a % on it, especially during major invasions... The only big problem for EB is that the engine doesnt really represent the situation very well. Although bizarrely at least one of the custom battles in the RTW Alexander campaign does it rather excellently. the biggest issue being that the AI just resorts blindly to one massive frontal assault with its best units or sits on top of the nearest hill, not the opposite bank while defending.

Personally in my fevered imaginings I would like to see siege type options brought up at these points. ie if the defender has been there a while he can fortify the position a little (although that option would be good anywhere on the map. Im not talking about actual forts because then you have no option to withdraw and just get stuck in them and cant get out without sallying, which is a nonsense IMO.) Also the attacking side would be able to build a pontoon bridge or rafts to cross the river elsewhere with certain units etc etc. Real life rivers are often far wider comparatively than represented in the game and in many cases could not be crossed on foot at all.

[stupac]

Well, historically, a great many battles were fought at river crossings. In fact it would be interesting to get a % on it, especially during major invasions... The only big problem for EB is that the engine doesnt really represent the situation very well. Although bizarrely at least one of the custom battles in the RTW Alexander campaign does it rather excellently. the biggest issue being that the AI just resorts blindly to one massive frontal assault with its best units or sits on top of the nearest hill, not the opposite bank while defending.

Personally in my fevered imaginings I would like to see siege type options brought up at these points. ie if the defender has been there a while he can fortify the position a little (although that option would be good anywhere on the map. Im not talking about actual forts because then you have no option to withdraw and just get stuck in them and cant get out without sallying, which is a nonsense IMO.) Also the attacking side would be able to build a pontoon bridge or rafts to cross the river elsewhere with certain units etc etc. Real life rivers are often far wider comparatively than represented in the game and in many cases could not be crossed on foot at all.

True enough, I remember in reading the Conquest of Gual, Caesar had to build bridges for his 2 invasions over the Rhine into Germania. Though I'm sure there were natural fords on many rivers, they may not have been suitable for moving large numbers of troops, not too mention baggage usually carried by cart, and many rivers, like the Rhine were just too big period.

[Foot]

Personally in my fevered imaginings I would like to see siege type options brought up at these points. ie if the defender has been there a while he can fortify the position a little (although that option would be good anywhere on the map. Im not talking about actual forts because then you have no option to withdraw and just get stuck in them and cant get out without sallying, which is a nonsense IMO.) Also the attacking side would be able to build a pontoon bridge or rafts to cross the river elsewhere with certain units etc etc. Real life rivers are often far wider comparatively than represented in the game and in many cases could not be crossed on foot at all.

This is absolutely impossible. Only us to modify the AI does not allow us to add new features to the battlemap. The closest you can get is the stakes that archers can setup before a battle, but we cannot automatically add them or stop an attacker from using them. We certainly cannot add anything you mention above.

Foot

[a completely inoffensive name]

The EB team can only improve upon what the engine will allow, not create entirely new stuff for it.

[russia almighty]

Yeah, I wish people would realize that.


Which, that is a good topic for one of you EB members to post. Saying, we can only work within the M2TW and RTW engines (and posting limits) would probably stop a lot of pointless topics.

[MarcusAureliusAntoninus]

Even if the limits were posted and stickied, with outlines that said what could and couldn't be done, people wouldn't read it and just ask the question. Though it doesn't matter, we'll keep saying when things can't be done. Just don't take it personally when someone answers a good idea with "hardcoded".

[Theodotos I]

I checked things out with Byg over on the Stainless Steel Forums at TWC, and he said that his Grim Reality AI mod would be compatible with any mod of MTW2:Kingdoms. Any chance that that would work in the ancient era? Just wanting to be helpful. :2thumbsup:

[Foot]

We will be looking to adapt the AI (particularly the campaign) to replicate the peculiarities of Classical diplomacy and warfare. An old member has rejoined us for just that purpose and the proposals so far look wicked. We may use other AI mods for information and research, but we will essentially be looking to modify the game in this way on our own terms.

Foot

[QuintusSertorius]

There's only two things that need "fixing" in the battle AI as far as I'm concerned. If they're actually possible to sort in MII:TW, then I'll consider EB2 a must have game, and worth getting a new computer for. Because there's no way I can run MII:TW with my current machine, and I can't upgrade it either.

Firstly, make the bloody AI capable of holding a line. Nothing fancier than not changing the formation once the battle starts. I've lost count of how many battles have turned into these ridiculously easy affairs because the AI marches it's units out one at a time, or sends one from a flank out to be beaten individually, or just sends them in all directions. Phalanx armies are the worst for this, but it happens with everything. This coupled with the issues with campaign AI are burning me out on RTW right now.

So I don't know if that's the kind of level it's moddable at, but make the AI move all it's units together, keeping the formation it started with, and I'll be very happy.

Second, stop the bloody generals killing themselves. Every single battle it seems, the general charges the front of a unit of formed-up spearmen, and gets killed. Often before the rest of his army has even engaged. So he gets killed and the rest of his army doesn't stick around for long.

Campaign-wise, anything that would make the AI actually combine it's troops into proper, balanced forces would be good.

And something that tells it when to stop attacking, particularly if it has few troops left. I know some of that is an issue with the military AI ignoring the diplomacy AI (cheers, CA!).

[overweightninja]

Firstly, make the bloody AI capable of holding a line. Nothing fancier than not changing the formation once the battle starts. I've lost count of how many battles have turned into these ridiculously easy affairs because the AI marches it's units out one at a time, or sends one from a flank out to be beaten individually, or just sends them in all directions. Phalanx armies are the worst for this, but it happens with everything. This coupled with the issues with campaign AI are burning me out on RTW right now.


I didn't play vanilla MTW2 for very long, but I'm playing Stainless Steel atm and imho the AI does a much better job of forming a cohesive body of troops if it has the numbers, not sure if the Stainless Steel team have modded the AI or not but if so I'm sure the EB can do just as good a job :2thumbsup:

[The General]

There's only two things that need "fixing" in the battle AI as far as I'm concerned. If they're actually possible to sort in MII:TW, then I'll consider EB2 a must have game, and worth getting a new computer for. Because there's no way I can run MII:TW with my current machine, and I can't upgrade it either.

Firstly, make the bloody AI capable of holding a line. Nothing fancier than not changing the formation once the battle starts. I've lost count of how many battles have turned into these ridiculously easy affairs because the AI marches it's units out one at a time, or sends one from a flank out to be beaten individually, or just sends them in all directions. Phalanx armies are the worst for this, but it happens with everything. This coupled with the issues with campaign AI are burning me out on RTW right now.

So I don't know if that's the kind of level it's moddable at, but make the AI move all it's units together, keeping the formation it started with, and I'll be very happy.

Second, stop the bloody generals killing themselves. Every single battle it seems, the general charges the front of a unit of formed-up spearmen, and gets killed. Often before the rest of his army has even engaged. So he gets killed and the rest of his army doesn't stick around for long.

Campaign-wise, anything that would make the AI actually combine it's troops into proper, balanced forces would be good.

And something that tells it when to stop attacking, particularly if it has few troops left. I know some of that is an issue with the military AI ignoring the diplomacy AI (cheers, CA!).

Agreed.

Oh, and I still posit the opinion that when it comes to bridge battles, less is better. I hate mindless slaughter in all its forms, and never station troops permanently at river crossings etc. It's okay to take advantage of them every now and then, especially if you're outnumbered/-classed, but to permanently, turn after turn to rape the enemy at river crossings... Sounds way too monotonous to me - not to mention like cheating (what kind of an enemy would just mindlessly try to push year after year somewhere where it gets nothing but completely annihilated each time, till the end of time?)

Oh well, I suppose my love for maneuvering (and micromanaging) just isn't suited for river/town battles. <.<

[QuintusSertorius]

Is there anything in the mod-able campaign AI around which direction a faction tries to expand? Anything that would stop Hayasdan trying to become the new king of the steppe would be welcome.

[Tellos Athenaios]

It appears there is...

[Foot]

well it ain't exact, but it might be able to help.

Foot

[Ludens]

Have you tried the win-conditions mod (https://forums.totalwar.org/vb/showthread.php?t=86313)? IIRC Konny updated it to EB1.1.

[QuintusSertorius]

Have you tried the win-conditions mod (https://forums.totalwar.org/vb/showthread.php?t=86313)? IIRC Konny updated it to EB1.1.

I'll give it a try.

[a completely inoffensive name]

It's been a couple months, I just wanted to know if much progress had been made in terms of the AI, or if it has still been lower on the to-do list while you create units.

[Puupertti Ruma]

Well, I'd quess it is quite impossible to test AI if there weren't no units to test with...

[a completely inoffensive name]

All you need is one maybe two units to test out the AI. If there is a phalanx unit that is finished technically you can make an entire army with those and try out the AI phalanx tactics.

[Megas Methuselah]

Give the EB Team a break, ACIN. They'll get down to testing when they will; most of them are probably writing their finals right now, much like myself. And trust me, the extra stress is not welcome.

[a completely inoffensive name]

I am not pressuring them, I was asking a question and Ruma suggested it was unreasonable, and I am just saying that it isn't.

[sirmic]

I would like to suggest to EB team to look into XBAI (XAI) from Xeryx team. I know that there are some questions (with Lusted) about functionality of all of code which Xeryx made up from the scratch, but i personally tried XBAI(battle ai) in few mods like stainles steel and even in vanilla brittain campaign and the result was breathtaking for me: Generals and other arent suicidal but tries to get your archers or charge into your heavy sword inf and on the other side ai spearman goes for your your cavalry, skirmishers support theirs infantry from back and masakres your engaged heavy inantry with their armorpiercing javelins(damn). Ai archers supporting from back too of course but when I try to flank ai and they have no unengaged troops then archers tries to stop me in desperate move and tries to buy a little time for his army (like i do when situation goes desperate...).
XBAI is portable to any mod - as the changes of the files related to BAI are made in universal matter.

Some another benefit is that Xeryx seems to be very active and open to help and answer any questions.

In recent time Xteam presented XAI -battle and campaign ai in newest build of darth ages, i am playing now and i am justified with diplomacy too.
Of course for more info take a look into Xeryx team forums i f you would like.

[a completely inoffensive name]

I think the EB team has said, they will make their own AI from scratch. I could be mistaken.

[sirmic]

Yes, it will be great to have AI(cai and bai too) from EB Team made from scratch.
I just suggested that Xeryx Team(+Darth and Condo) already put a lot of effort into developing of new AI and at least an inspiration from their work could be for good of EB2.

So I am interested if EB team members actually know about project XAI (which is in stage of 3.2F).

I write all of this with best intentions towards ours gaming eperience. What is most important on strategic game? The challenge which you have to face from your oponent. Playing strategic games is for me enjoyable like playing sachs and i like to compete with very able oponent... in sachs dveloprs of AI already made big achivements..:2thumbsup:

[Cullhwch]

Agreed.

(what kind of an enemy would just mindlessly try to push year after year somewhere where it gets nothing but completely annihilated each time, till the end of time?)
<.<

General Cadorna would like to have a word with you...