A Better AI.

[Meatbuster]

...just shortcutting by giving the AI express technology just takes the fun out of it.

So I take it that are you accusing Blake or Firaxis of giving free techs to AI?

In my recent unfinished game (Large Monarch Pangaea 2.08 with Blake 11-07) as Augustus Caesar (read: not Financial) I find that the only way to keep up is to have friends, preferably Financial ones. I was able achieve Military Tradition in 1300's, but I ignored a lot of techs. Mainly the Feudalism-Guilds-Gunpowder line (which I traded for from my best buddies Mansa and Wang). But before I made those friends, I was fighting Macemen with Praetorians... of course that sucks!

edit: I win by diplomacy after signing defensive pacts with 3 civs and converting another to confucianism.

 

[kettyo]

While I empathize with everyone on subtle AI improvements like positioning of cities etc, there seems to be a fundamental flaw in the AI which has been referred to a number of times in this tread but seems to have not gotten a lot of traction – how is it fun to play an AI that is screaming ahead of you in technology.

I was/am a regular Monach player, after loading the patch (2.08) I am now facing Cavalry in the year 1280 AD – one minute the AI had Knights 2 turns later in 1280AD it has Cavalry. This to me is nonsensical. At this rate the AI will launch it’s spaceship in the 1700s. Prior to the patch Cavalry would be obtained somewhere between 1500 and 1600 on this level. What happens on the higher levels? Rocketry by 1000AD?

Perhaps I could manage my cities a little better, perhaps I could pick city locations a little better, there are probably infinite things I could do better, but I don’t see anyway I can shave 200-300 years off research and be competitive in 1200 AD against Cavalry.

Even going to war is suspect when your mace men and catapults are competing against units of strength 15.

You can make it harder for the Humans, you can smarten the AI….. just shortcutting by giving the AI express technology just takes the fun out of it.

The goal of Blake would be to make the AI competitive in lower levels too not on Monarch where they get lots of bonuses against the human.
In ideal case they would be competitive in Noble where the rules favor the AI and human roughly the same.

 

[kettyo]

The AI forces not pillaging much now i don't know if it's a good idea.
They often suicide attack strong cities now.
It would be better for them to pillage everything instead.
I suppose Firaxis made them so pillage crazy because the AI can hardly make an organized attack on a well defended city.
Also i think they should use much more big stacks instead of individual units or small stacks of 2-3 units. They usually start the war with an invading stack if they attack but then don't muster new stacks.
Attack in large numbers in the later stages of the war too would cerainly help their offensive a lot.

 

[Big Roy]

I'm not accusing blake or anyone else of just butchering the game. I'm pleased that someone is taking an active interest.

Cavalry in the 1200s just doesn't seem realistic. What happens on Immortal - 1AD? How can we compete? You can strengthen AI by making it harder for the human player - increase cost of tech - I enjoy the early/mid game a great deal but find it over very quickly - especially now.

 

[analyst]

I'm not accusing blake or anyone else of just butchering the game. I'm pleased that someone is taking an active interest.

Cavalry in the 1200s just doesn't seem realistic. What happens on Immortal - 1AD? How can we compete? You can strengthen AI by making it harder for the human player - increase cost of tech - I enjoy the early/mid game a great deal but find it over very quickly - especially now.

You can compete by lowering your difficulty level until you can handle the AI now making more intelligent decisions.
The higher difficulty levels are there if you can already handle the lower ones and you're bored with them. If you can't handle Monarch, go down to Prince. The whole purpose is to have a challenge; perhaps Prince will now be a challenge to you.

 

[Big Roy]

Is there a broken record? - Stock standard response - drop your level.

You're missing the point, it is not realistic! Why don't we just move it to another world where gunpowder and computers exist just after the birth of Christ.

You can make it more difficult for the human players and still maintain a sensible time line - get it?

 

[Thedrin]

While I empathize with everyone on subtle AI improvements like positioning of cities etc, there seems to be a fundamental flaw in the AI which has been referred to a number of times in this tread but seems to have not gotten a lot of traction – how is it fun to play an AI that is screaming ahead of you in technology.

I was/am a regular Monach player, after loading the patch (2.08) I am now facing Cavalry in the year 1280 AD – one minute the AI had Knights 2 turns later in 1280AD it has Cavalry. This to me is nonsensical. At this rate the AI will launch it’s spaceship in the 1700s. Prior to the patch Cavalry would be obtained somewhere between 1500 and 1600 on this level. What happens on the higher levels? Rocketry by 1000AD?

Perhaps I could manage my cities a little better, perhaps I could pick city locations a little better, there are probably infinite things I could do better, but I don’t see anyway I can shave 200-300 years off research and be competitive in 1200 AD against Cavalry.

Even going to war is suspect when your mace men and catapults are competing against units of strength 15.

You can make it harder for the Humans, you can smarten the AI….. just shortcutting by giving the AI express technology just takes the fun out of it.

I'm not accusing blake or anyone else of just butchering the game. I'm pleased that someone is taking an active interest.

Cavalry in the 1200s just doesn't seem realistic. What happens on Immortal - 1AD? How can we compete? You can strengthen AI by making it harder for the human player - increase cost of tech - I enjoy the early/mid game a great deal but find it over very quickly - especially now.

I'll highlight the parts which are causing peope to ask you to drop a level.

a regular Monach player

What happens on Immortal

The difficulty levels were designed with the old AI in mind, not the new one. If you play on Noble and still have a noticeable problem with the timeline then you may have a valid complaint.

 

[player1 fanatic]

Is there a broken record? - Stock standard response - drop your level.

You're missing the point, it is not realistic! Why don't we just move it to another world where gunpowder and computers exist just after the birth of Christ.

You can make it more difficult for the human players and still maintain a sensible time line - get it?

It was always like that, but on higher difficulty levels.
With Blake AI it just happens earlier, since AI is smarter.

Drop the level...

 

[asabahi]

Is there a broken record? - Stock standard response - drop your level.

You're missing the point, it is not realistic! Why don't we just move it to another world where gunpowder and computers exist just after the birth of Christ.

You can make it more difficult for the human players and still maintain a sensible time line - get it?

Perhaps the tech tree needs to be tweeked a bit. I think I understand what you are saying. I used to win on prince regularly but after the patch I noticed that even if I went to war early and did what I used to do before, I was behind. So I adapted a bit and now I am winning again (not through waring all the time). The AI can sling shot a bit past a player in the middle ages but it falls behind again midway through the industrial age. I believe this is do to players taking better adavantage of biology, ie assigning more specialist, or cottaging more for extra cold hard cash. Also I have learned to tech trade much more aggresively than before even if the trade is a bit one sided.

 

[kniteowl]

Perhaps the tech tree needs to be tweeked a bit. I think I understand what you are saying. I used to win on prince regularly but after the patch I noticed that even if I went to war early and did what I used to do before, I was behind. So I adapted a bit and now I am winning again (not through waring all the time). The AI can sling shot a bit past a player in the middle ages but it falls behind again midway through the industrial age. I believe this is do to players taking better adavantage of biology, ie assigning more specialist, or cottaging more for extra cold hard cash. Also I have learned to tech trade much more aggresively than before even if the trade is a bit one sided.

Interesting...

Hey Blake does the AI know how to take advantage of biology? And make an Early Beeline towards if it gives them an Advantage?

Eg- Biology = More Food = More Pop = More Producitve Tiles Being Worked

 

[Wodan]

Ok let me see if I have this right. People complain that they're playing on high levels and the AI are getting techs real early in some games, such that "gunpowder and computers exist just after the birth of Christ".

Yet the reason they're playing on high level is because their game style exploits the AI weaknesses, especially lackluster war preparation and war execution.

Is not a reasonable solution to ask Firaxis (and Blake) to tweak the AI such that it is better at war? Would not the inevitable result be that the above people drop their level, thus the AI will simply not have the capability to have such huge tech advances?

Wodan

 

[Thyrwyn]

Some comments on the new AI:

1) allow me to add my "thank you, Blake" to the chorus.

2) I've noticed that the AI tends to build more workers earlier than it used to, which is great. I have also noticed that the AI builds more cottages - great for them, better for us when we conquer them

3) the AI settles much more aggressively in the land grab phase than it used to, which is also great. It also means that early wars are much more necessary. I started a game last night (as Brennus), founded a religion and settled a second city. The Shaka settled two cities that isolated me in a corner of our landmass. He started to my south and each successive city was north and east of his previous territory. Fortunately, I had two clams, pigs, lots of trees and 3 hills near my capitol so I was able to build up an Army and go to war early to get out of my corner.

4) Trade: Blake mentioned earlier (or elsewhere) that one of the effects of his improvements was the increase in the trade-ability certain techs (such as Compass). The one big change I have seen is that the AI no longer has oodles of cash lying around waiting for the player to come to market. for whatever reason (I'm guessing deficit research), the AI in the games I've played have very little cash on hand, unless you catch them after a failed wonder bid. This has had a huge impact on the value of trade bait across the board.

I used to be able to count on getting a couple of techs that I had ignored from 1 or 2 of the AI, then cleaning out the treasuries of the rest to fund future research - now there are no treasuries to clean out!

The new AI makes for a very different game experience. Thanks again!

 

[Arlborn]

4) Trade: Blake mentioned earlier (or elsewhere) that one of the effects of his improvements was the increase in the trade-ability certain techs (such as Compass). The one big change I have seen is that the AI no longer has oodles of cash lying around waiting for the player to come to market. for whatever reason (I'm guessing deficit research), the AI in the games I've played have very little cash on hand, unless you catch them after a failed wonder bid. This has had a huge impact on the value of trade bait across the board.

I used to be able to count on getting a couple of techs that I had ignored from 1 or 2 of the AI, then cleaning out the treasuries of the rest to fund future research - now there are no treasuries to clean out!

The new AI makes for a very different game experience. Thanks again!

Yes! This is such a big change! Byebye free cash by smart tech trades, they never have more than 50 gold >.< But one thing that Im wondering about that change, does the AI have money enough to upgrate units as it used to be or they burn it all in science? If not this can be a downside...

Thanks again

 

[asabahi]

I think certain civ do use their income to upgrade troops. When playing a game this weekend (prince), i was far ahead in power, and then Mehemds power rating spiked ahead of me in one turn. This was a huge jump and he had less than half the cities I had. The only way to explain this is a massive upgrade since we had tech parity.
I like the changes to the AI. I can no longer sit on my behind after getting a lead early and milk the rest of the game for points. Have to stay on top of it and try by the use of diplomacy or war to slow down the civs that are pulling away.

 

[·Imhotep·]

Blake, I've played some games now with the v2.08 Warlords - and hell, you did a good job ! I was regularly winning on Prince before, but now Prince is a real challenge again (and I thought that I could move up to Monarch some time - forget it !). Really crashing is a game I'm playing right now - I'm playing on Emperor tag-teamed with Washington. Well it's natural that the AI is rocketing away technology-wise with the aid of the human player - but more astounding to me is that Wang Kon keeps up with us in tech. And he has significantly less landmass to use than we do... Just amazing !

Imhotep

 

[civzombie]

"ut one thing that Im wondering about that change, does the AI have money enough to upgrate units as it used to be or they burn it all in science? If not this can be a downside."

Heh, a little off topic, but I always like scraping away all a civ's available money right before I declare war on them. I try demanding tribute, and if they don't pay, ill sell them a non-military tech on the cheap just to raid their coffers. They think they are getting a deal, and then wamo!

On topic, I have previously thanked Blake, but I thank him again. Rome is kicking my ass

 

[kettyo]

It happened again.

Alexander attacked Catherine (for the second time in the game) with a stack of 5 units and no reinforcements.
I remember with the default AI they attack with large stacks at least i remember Brennus attacking Mansa Musa with a HUGE stack of horse archers, taking the border city instantly and devastating a lot.

The new AI maybe move AI's to attack with fewer forces?
If this is the case i think it should be changed back and when they lose local superiority then tell AI troops not to suicide attack cities but pillage instead.

Otherwise i find your mod great, Blake!

Hope we see you soon.

 

[ABigCivFan]

Ketto, it is not a great idea to have the AIs send HUGE stacks to attack. This is FATAL to the AIs because human players can effectively use siege weapons to devastate this stack (think about a single 40 hammer catapult can damage a stack of 5-6 110 hammer rifleman and reduce their strength by 10-15%! what a bang for the buck.).

Big Roy, let me share my experience on my current game: Immortal/Standard/Normal speed as Cyrus. Yes, the AIs research at a blazing rate; in the early part, I captured 4 German cities since Frederik only had Archers. That completed my land grab phase (8 cities). I was able to stay in the tech race by aggressive trading. Alphabet, metal casting netted several techs and some cash, Campass gave me a real boost in techs and gold as the AIs loved it.

I was on good terms with everyone. Trading anything and everything even for just 150 gold. I found that if I don't trade my techs, 2 turns later everyone will have them anyway so thats better than nothing.

i build walls/castles in my border cities and have a solid defence force while teching and growing. The AIs did pull away around Education when EVERYONE refused to trade and gave me "We fear that you become too advanced" crap. I had to research almost everything myself.

Julius was the most powerful, I converted to his religion to suck up on him, used organized religion. Had open border with everyone (except Toku). I "visited" everyone with immortals. Everyone had mostly defensive units except Julius and Ragnar (the 2 leading scores). So I visited their cities regularly.

They got Rifle ~1100 AD, and Calvary ~1450 AD.

After they had rifleman(I was still researching gunpowder) and started massing knights, I could no longer sit around and hope nothing happens. I bribed Julius(with education) the leading score to attack Ragnar (2nd score). I had units near the action and witness the fight. Julius sent packs of 2 calaries to pillage, very effective. 10 turns earlier. Ragner was foolish enough to sent a large force across the map to fight Asoka. Now he was scambling to defend.

Julius's pillaging parties reduced some of Ragner's territory to 4000BC like, and attack cities with stacks of 10+ Rifle/calvary/Trebs. I was impressed by the AI combined army actions and pillaging. Julius also attacked on 2 directions simultaniously in north and east, capturing 3 cities at the end.

You mentioned AI unit upgrading, yes, the upgrading is immediate. I don't know where they get the cash, but I saw Julius upgrade 20+ war elephants and knights to Calvary in one single turn. Maybe the cost is much lower for them at this level, and they paused research to get the cash. Julius did get a lot of gold from pillaging Ragnar's towns apparently casue during the war, he had an excess of 1000+ gold every turn!. But this is realistic because human players do exactly the same. You can not rely on old units during the peach time, you have to continuously build units to deter AI invasion. Also when the war comes, you HAVE to pause research and upgrade almost every unit since you WILL be short on the latest units.

While they were busy at war, I frantically researched to rifling and next to Mil Trad. It gave me some time to breath, I also massed a large number of knights for upgrading. I plan to immidietely fight Julius as I get Mil Trad and upgrade all my knights cause he was getting too strong and pulling ahead in techs. And I will bribe anyone who could help me.

So higher levels are definitely playable and much more exciting when everying decision, every turn, every trade matters, and lots of micro-management. You have small windows of opportunity, mostly involve manipulating the AIs diplo style and smart waring. I check my F4 screen almost every turn for possible trades and bribes. It was a pain but had no choice.

I am now (1550 AD) 2nd in pts, 2nd in populations and have the largest land. but trailing in techs and way behing in troop count. I am 2 techs away from Mil Trad. 7-8 turns to research that tech(i was doing 70% science and 500 beakers/turn). I am not sure if I could win against Julius, but would try my best.

My battle plan for my only chance:

1. Before Julius gets infantry
2. Mass Pinch and formation Calvaries, pinch rifles, catapults(possibley canons if i get there)
3. Bribe someone to fight him with me; even better bribe Julius to fight someone far away(diverse his forces) and back stab him. He is pleased with me since I have done well sucking up to him the whole game(religion, one-sided trades and etc).
4. Bait him to attack my border cities and fight him in MY territory.
5. Concentrate my forces to destroy his initial SODs by using many kamakazi catapults and my elite offensive units.
6. The winning or lossing of most of these games are decided in these key single battles. If the AI losses his initial large forces, he will never be able to catch up to the human in Military. After you defeat the leader on the chart, you will rule the rest.

 

[Arlborn]

Ketto, it is not a great idea to have the AIs send HUGE stacks to attack. This is FATAL to the AIs because human players can effectively use siege weapons to devastate this stack (think about a 40 hammer catapult can damage a stack of 5-6 rifleman and reduce their strength by 10-15%!).

Big Roy, let me share my experience on my current game: Immortal/Standard/Normal speed as Cyrus. Yes, the AIs research at a blazing rate; in the early part, I captured 4 German cities since Frederik only had Archers. That completed my land grab phase. I was able to stay in the tech race by aggressive trading. Alphabet, metal casting netted several techs and some cash, Campass gave me a real boost in techs and gold as the AIs loved it.

I was on good terms with everyone. Trading anything and everything even for just 150 gold. I found that if I don't trade my techs, 2 turns later everyone will have them anyway so thats better than nothing.

i build walls/castles in my border cities and have a solid defence force while teching and growing. The AIs did pull away around Education when EVERYONE refused to trade and gave me "We fear that you become too advanced" crap. I had to research almost everything myself.

Julius was the most powerful, I converted to his religion to suck up on him, used organized religion. Had open border with everyone (except Toku) "visited" everyone with immortals. Everyone had mostly defensive units except Julius and Ragnar (the 2 leading scores). So I visited their cities regularly.

They go Rifle ~1100 AD, and Calvary ~1450 AD.

After they had rifleman(I was still researching gunpowder) and started massing knights, I could no longer sit around and hope nothing happens. I bribed Julius(with education) the leading score to attack Ragnar (2nd score). I had units near the action and witness the fight. Julius sent packs of 2 calaries to pillage, very effective. 10 turns earlier. Ragner was foolish enough to sent a large force across the map to fight Asoka. Now he was scambling to defend.

Julius's pillaging parties reduced some of Ragner's territory to 4000BC like, and attack cities with stack of 10+ Rifle/calvary/Trebs. I was impressed by the AI combined army actions and pillagin. Julius also attacked on 2 directions north and east, capturing 3 cities at the end.

While they were busy at war, I frantically researched to rifling and next to Mil Trad. It gave me some time to breath, I also massed a large number of knights for upgrading. I plan to immidietely fight Julius as I get Mil Trad and upgrade all my knights cause he was getting too strong and pulling ahead in techs. And I will bribe anyone who could help me.

So higher levels are definitely playable and much more exciting when everying decision, every turn, every trade matters, and lots of micro-management. You have small windows of opportunity, mostly involve manipulating the AIs diplo style and smart waring. I check my F4 screen almost every turn for possible trades and bribes. It was a pain but had no choice.

I am now (1550 AD) 2nd in pts, 2nd in populations and have the largest land. but trailing in techs and way behing in troop count. I am 2 techs away from Mil Trad. 7-8 turns to research that tech. I am not sure if I could win against Julius, but would try my best.

Can you post a save? I would like to give a look on this

Sorry for the offtopic..

 

[kettyo]

Ketto, it is not a great idea to have the AIs send HUGE stacks to attack. This is FATAL to the AIs because human players can effectively use siege weapons to devastate this stack (think about a single 40 hammer catapult can damage a stack of 5-6 110 hammer rifleman and reduce their strength by 10-15%! what a bang for the buck.)

I admit you're right on this just i've seen two examples of suicide war declarations by the AI in a short time.
Ragnar attacking me with 2 ! spearmen
And Alexander attacking Catherine with a 5 unit stack.
And no reinforcement stacks were sent by the attackers.
It would be best to tweek the AI to muster stacks while in the war too, if it's possible.