GENERAL: Civ 3 vs. CtP2

[Gilgamensch]

Reminds me of something somebody already mentioned:

The AI doesn't need to be tought to defend cities, but rather points/borders.

What is the need to place in every town 12 units, if you could secure your country with only one 12 stack on the border on a mountain in a fort?

Best example for stupid AI behaviour for this: CIV 1(&2&3&XXXX)

Place a good defensive unit on top of a mountain and it keep attacking like hell.............

(Before CIV players are bashing me, that is typical for 'any' computer game....AFAIR)

[Peter Triggs]

It might cheat in the fact that those units attack your weakest city first...

This is the kind of cheatings that I don't want to see in a CtP2 mod, this is an awful one and what I call a spoiler.

IIRC, in vanilla CTP2 the AI is actually programmed to go after your weakest cities first. The difference between it and Civ3 is that the CTP2 AI will only go after the cities that it knows about and then only if it doesn't have a better target.

In Civ3, it's said that the AI knows the whole map. So if it's programmed to prioritize attacks on weakly defended cities, this must be really annoying.

In Call to Conquest, I added a 'raid' goal so that the AI would occasionally send out a raiding party into your interior. This was meant to keep you on your toes but not to be a primary AI goal.

[The Big Mc]

but it is easy to see which city is least likely to be heavily defended even so you should occasionally drop something and attack a presumed week city we a battle damaged stack

[Gilgamensch]

Originally posted by The Big Mc
but it is easy to see which city is least likely to be heavily defended even so you should occasionally drop something and attack a presumed week city we a battle damaged stack

How can you see if a city is weekly defended?

The only cheat, which the humna player can use is the drag-cheat, which I personally hardly use.

[centrifuge]

Originally posted by child of Thor
Dales work in the WAW mod produced a decent Enemy AI attack capability, it was short in the all round game plan i think(city management etc?), but the way the AI handled stacks and what they consisted of seemed a huge improvement.
I'm sure Dale,Martin,Peter,Jbytheway and others have many ideas we can use to get the right balance and feel for the AI.

Yep WaW's AI was excellent at attacking, but the defense of the cities was horrific. In newer versions this may have been changed... but what my game plan turned into was: try to stay alive long enough to sneak a few tanks behind the front lines and take some of their undefended cities...

[centrifuge]

Just another suggestion, Unless significant changes are made to the AI, let's change the "Impossible" level to something more appropriately titled

[The Big Mc]

like very very easy

and yes it is you generally know a city of pop 2 is less defended then a city say 50

[hexagonian]

Solver hits the nail on the head with his analysis of the differences...

Some observations regarding what I have seen in my Cradle Modwork.

The basic CTP2 AI cheats are not all that drastic - in fact, in the vanilla game, I felt that they were less drastic than in civ3. Nevertheless, they have been greatly bumped up in the Mods.

Cradle gives the AI free unit upgrades as opposed to the human player having to pay upgrade costs. (Actually, if I recall correctly, doesn't the AI in the higher levels of civ3 get free upgrades???) This is not such a big deal to me because there is a lot of gold that can be generated in the game. The end result is that a player has to make more choices in what to do with his surpluses, since that gold serves for no other purpose other than RBing in the vanilla game.

I do think that the most current version of Cradle has been able to get the AI to build more cities (one of my major gripes with earlier versions of the game), but this has been done by alterations in strategies.txt and is not a cheat. Simply a matter of adjusting the priorities for settler construction and loosening the restrictions on where the AI will build cities. In my last game, I went into cheat mode to see how many cities each civ had and they were all near the cap for their respective government (at that time, 15 per).

The most powerful cheat in my Mod, IMO, is the PW/Gold SLIC file. This file adds a bonus to the AI based on population in gold and in PW funds. The end result is that the bonus snowballs in its benefit. As the population grows (and it does because PW will speed up growth), more PW is added - and more population means more gold, more production, and even more science. Add this to the normal cheats and yes, the end result is that I feel the (Modded) CTP2 AI does get a lot more bonuses than civ3.

Nevertheless, this cheat is an add-on and can be disabled, and I would not be opposed to removing it. In fact, I would love to eliminate it, but for the following...

The main reason I have it in Cradle is because of the lack of focus on AI attacks. This cheat helps the AI maintain an advantage in unit strength and army size to offset its deficiency in conquest focus. In earlier versions of Cradle, the PW/Gold SLIC helped offset the fact that the AI did not expand as well as it could, but I've found that in versions 1.34-1.35, with the new emphasis on settler expansion, the AI gets a much larger lead because it now peacefully builds more cities.

I definitely agree with Solver that this area of conquest is one of the weakest areas in CTP2. At least in civ3, the AI seems to have a definite focus when it is attacking, and civ3 AI/AI and human/AI alliances do work much better. I do fear the AI in civ3 more than the AI in CTP2 because it is more coordinated.

The Frenzy SLIC does help the CTP2 AI to send stacks within your territory and attack cities, but it also prevents the AI from maintaining garrisons in those cities that it does capture. The end result is that you expend a great deal of resources to take back those cities (generally you have to rebuild buildings in those cities, as well as replace destroyed units) - and having to deal with multiple AI stacks within your borders is a huge challenge - try coupling your defensive needs with your desire to expand via conquest. Sometimes, the AI will cut through several of your cities too, but it doesn't have a coordinated feel to it.

Nevertheless, the Frenzy code is not a cheat and I tend to play Cradle with the following restriction - I cannot disband captured cities - which ends up making it harder to completely remove a civ from a game. The end result is more along the lines of history too.

My main gripe with the civ3 diplomatic model is that it generally forces the player to slavishly trade for techs, negating the need to develop your own tech. This is the main reason why the civ3 AI is able to stay competitive, because it is such a tech whore.

To some degree, the difficulty in maintaining diplomatic and alliance ties in Cradle is realistic for the timeframe that is in Cradle and it does make it harder for the player to mount an effective offensive because you cannot ever leave yourself weak in any area of your empire. It does not bother me that diplomacy in the CTP2 model is usually harder and less predictable - generally the fact that you do not know if a friendly civ will suddenly turn on you mirrors human nature.

OK, so what am I aiming at with all of this...
I want an alternative to Frenzy. Some sort of alteration that will give the AI more focus on conquest in an efficient manner.

Generally, I would like the AI to maintain an aggressive stance toward the human player as the default status - most importantly, I would like the AI to place a priority on holding onto those cities it does capture. This would allow for the need for artificial bonuses to be greatly reduced. I am against making tech trading easier, but I wouldn't mind seeing more trustworthiness in AI/AI and human/AI diplomacy, at least along the lines of maintaining peace and it should be costly to maintain a peaceful state between civs if there is a great difference between the two in the rankings.

[Gilgamensch]

Actually hex, after reading your comment:

One thing I would like to see implemented: If the AI was first to discover and build certain key-techs (Knights/Infantrie/....) that it would alter to a more attack style, as it also happened in reallity. Whoever was first with a modern way to attack, normally did so.

[Solver]

Great post Hex . And I must agree that, indeed, any kind of PW cheating for the AI is pretty annoying, as it gets obvious, but it's the only way for the AI to stay competetive.

Frenzy is very good code. I have had decent stacks in my land with it - although I've also seen some AI units getting stuck between its own 12 stacks, heh. However, Hex and myself consider coordination to be of importance. I think that those of us who have played both Civ 3 and CtP2, will agree that Civ 3 AI attacks feel much more coordinated or focused.

My main gripe with the civ3 diplomatic model is that it generally forces the player to slavishly trade for techs, negating the need to develop your own tech. This is the main reason why the civ3 AI is able to stay competitive, because it is such a tech whore.

It's not a diplomatic model fault, but rather one of the major gameplay faults in Civ 3 - maybe even the worst one. On any level above Regent, you can't keep up with the AI research wise, no matter how hard you try. This is because of both the bonuses that AI gets, the cheap AI-to-AI trade, and its rather efficient management. So, the only way for you is to go 90% or 100% taxes, only researching the techs you know AI doesn't. This is as far as gameplay goes.

And now, diplomacy faults step in. If you have, say, Monotheism, a tech AI will rarely research, you can easily talk to all other civs in that turn, and get like 4 other techs and several hundred gold in return. This makes it very inefficient for you to do your own research, even if you can research at about the same speed that the AI can.

After midgame, human is usually in a position when he can finally start research of his own, and do it at a speed considerably better than the AI. And here you go on the next problem - you can sell those techs for everything that the AI has. Since it no longer has any other techs, it will offer you money - lots of it. A decent lump sum, and a nice gold-per-turn deal. The result: you get so much money from these tech trades that you can go 90-100% science, while the AI is spending almost all of its income each turn to pay you. This is why in the Industrial Age and later, the human ends up having several thousand gold, rapidly rushing infrastructure where needed, while the AI civs have under 100 gold.

[The Big Mc]

I know it is a bit cheating but wasn’t that the purpose of integrating a spy unit into civ you can steel tech not just try and buy your way to alpha century lets just get the ai to play sneakily that way it will keep up with tech and the ai can trade with itself so with one steel the hole ai opposition gets a help

I don’t remembered the ai ever saying they wanted a tech of me probably because I am a mind less conqueror or something

[Solver]

Stealing techs is the way I go about in CtP2 when someone has techs, because I can almost never get a deal with the AI.

[The Big Mc]

another problem i see with ctp is the bog standard ai does not produce and city imps i once moded it as an experiment in ctp1 the printing press so it gave a stupid amount of science it took the ai about 50 turns to make one by this time i was in the diamond constructing something good

ps any better ideas for age names then diamond

[Myrddin]

I like the effect in Civ3 of not being able to use roads in enemy territory - would this help the AI defend by giving them time to respond to player attacks?

The limit for units in each tile is sensible, but is there a way to turn it off in cities - so that you don't have to build by-passes to get armies through/around cities - may also help the AI pathing while at war

[The Big Mc]

yes it would help but it is totally unrealistic