What do you expect from the AI in Civ 4

[Deep_Blue]

That's not the question I asked. I wanted to know the difference between science fiction and tech that likely won't exist till 2035.

Same thing .. but I when I said:
"Some people consider it as science fiction . I think it is just a far away technology."
I meant some people think its impossible and fantasy . and I think it will be real one day.

 

[warpstorm]

Good science fiction is based on extrpolation of known facts.

 

[Deep_Blue]

((I suppose Science Fiction may be creative fun stuffs that are likely not possible... Where as distant tech is something within reach that just needs a lot more work and time before it will be Science Fact.))

Exactly

 

[bshirt]

Deep Blue;

Do you think there is a reasonable chance that CivIV's AI will be improved as much as CivIII's was from CivII?

While there are endless complaints, many justified, regarding the failures of CivIII's AI, to me, relative to CivII, it was HAL.

 

[Deep_Blue]

I cant tell before actually try the game, however there is a chance that the AI in civ 4 will be a success compared to civ3.
I dont know how much development effort Firaxis spent on the AI.
Because of the transition from 2D to 3D we can conclude that more effort was done on graphics rather than on AI. However there is a chance for a significant improvement in the AI if AI development was running in parallel with graphics development.
I hope they will bring an AI that is much better than civ3. We have to wait to see if they have achieved that.

 

[labguy23]

The major issue with the Civ AI is it's inability to plan ahead or access it's chances for success.

A civ with bowmen should inherently understand that is cannot win a war with a civ that has infantry or modern armor. However, that does not mean that it does nothing, instead it should focus on denial for resourses and technology (trade embargos). If they convince enough other civs to join them, it can have a dramatic effect on your civ's ablility to thrive.

It could even enlist the use of spies to try to level the playing field by stealing tech, production, or money. It could even try to create dissention in your cities as a way of slowing you progress. However, it would refrain from actual war until such time it could pose an actual threat to you.

In addition, if the civ recognizes that it has technology and military power on par with yours, it needs to intelligently choose targets (and stay focused on those goals over time without constantly changing objectives) and horde resources sufficient to have a resonable chance of success. In other words, the attacking civ would have to reach a critical mass of sorts before actually launching an attack (even though it may have decided to attack you 10-20 turns ago). This critical mass could be achieved through making alliances with other cultures that can rivil yours or simply amassing more forces themselves. Once critical mass is achieved, then it could launch it's attack trying to seperate you from you valuable resources, wonders, or core cities (attacking targets of opprotunity along the way).

The AI would need to not tip it's hand as well. A good player will prepare before the invasion, only revealing his intentions when he is ready to spring the trap. The AI needs to do the same.

Also, things like wonder creation, valuable resources, level of technology, military might, culture, etc should drive the AI bloodlust. Other civs should always seek to knock of the top dog (unless they are on good terms with them). This means you will need to cultivate a good relationship with all civs if you have a sizable advantage. If you exclude some civs and they are only nuetral to you, they will grow increasingly jealous of you, eventually forcing it to take diplomatic, military, or subversive actions against you.

 

[HourlyDaily]

A civ with bowmen should inherently understand that is cannot win a war with a civ that has infantry or modern armor...It could even enlist the use of spies to try to level the playing field by stealing tech, production, or money.
.

I can see it now, one civ stuck in the middle ages sends a spy envoy to a modern civ:

Picture one man in pantaloons and a wig wearing Groucho Marx glasses and moustache to dusguise himself from a General Patton type.

"Ahem, excuse me good sir, but could you please explain these big mechanical beast things to me."

 

[labguy23]

People in LA prolly wouldn't look twice at a guy like that.

"Hamlet must be playing at the local theater."

 

[Aussie_Lurker]

Actually Deep_Blue, the transition from 2D to 3D does not suggest more work on graphics than on the AI-for two very important reasons:

1) Within Firaxis, these different jobs are not a 'zero-sum game'. You have certain members of the team whose primary task is graphics, whilst another part of the team is almost completely devoted to things like the mechanics and AI. These two groups work largely in parallel, though I wouldn't be suprised if there was occasional cross-over.

2) I have it on very good authority that it is in fact easier to do graphics for 3D games than it is for 2D games, and that the graphics component of the game are usually done almost entirely in the last 6-8 months of a game's development cycle-wheras the AI has been consistently 'tweaked' from almost day 1.

Something which gives me hope for the AI in Civ4, though, is that they are apparently building it with an eye on what they have seen going on in multiplayer games done by Firaxis staff and the various playtesters. For example, they are seeing what 'exploits' players are trying to resort to, and designing the AI to effectively counter them. This means not only will we get a better MP game-because it was built from the 'ground up', but that the MP design could lead to a better SP game, as the AI has already been 'taught' to deal with some of the more 'ruthless' tactics of the CivFanatic !

Yours,
Aussie_Lurker.

 

[Melendwyr]

2) I have it on very good authority that it is in fact easier to do graphics for 3D games than it is for 2D games,

Bwa? I really can't imagine why that might be so.

 

[Meleager]

Bwa? I really can't imagine why that might be so.

Its my understanding (though I could be wrong) that in civ3 you had to create the graphics in 3d and then convert them to 2d. In civ4 you can take out the second step.

 

[Aussie_Lurker]

You have it in one, Meleager. According to those 'in the know' (Warpstorm, for instance, whom I have learned to trust implicitly) has indicated that 2D animations are almost always done in 3D first, then converted to 2D when implemented. Doing the game in 3D to start with removes this middleman step, thus reducing the overall time necessary to develop the graphics for the final game. Plus, as they clearly already have prior experience with 3D design tools and the like, its not as though the team needed to learn a whole new skills set.
Even were this not the case, though, my first point is still valid-that graphics and AI are almost certainly NOT a zero-sum game, as they probably have different people working on each element.

Yours,
Aussie_Lurker.

 

[Melendwyr]

Wow. What a waste of effort that must have been.

Perhaps that's why the graphic artists have persisted in keeping the unit's graphic size so large. After so many years of their work being covered up and mostly eliminated, they're dead set on having their beautiful units where they can be clearly seen. :c)

 

[Deep_Blue]

1) Within Firaxis, these different jobs are not a 'zero-sum game'. You have certain members of the team whose primary task is graphics, whilst another part of the team is almost completely devoted to things like the mechanics and AI. These two groups work largely in parallel, though I wouldn't be suprised if there was occasional cross-over.
Aussie_Lurker.

I dont know how Firaxis work , that why I said the AI could be improved if the work on AI and graphics was running in parallel.

2) I have it on very good authority that it is in fact easier to do graphics for 3D games than it is for 2D games, and that the graphics component of the game are usually done almost entirely in the last 6-8 months of a game's development cycle-wheras the AI has been consistently 'tweaked' from almost day 1.

I dont agree that 2d is harder than 3d in Civ. because in civ3 you only have to make animation for the units. while in civ4 you need to make 3d model for a world (containing mountains, ocean,...) + cities +units animation+ other.. there is a lot of work to do in 3D. So working on 3D is time consuming.
However I think they are using Pirates engine , this would simplify things but still require a lot of work.

 

[eaglefox]

do you think playing according to the civ specific abilities and traits will be the key to victory with that particular civ? more importantly, do you think this kind of play will increase the AI's chances of success?

 

[warpstorm]

I dont agree that 2d is harder than 3d in Civ. because in civ3 you only have to make animation for the units. while in civ4 you need to make 3d model for a world (containing mountains, ocean,...) + cities +units animation+ other.. there is a lot of work to do in 3D. So working on 3D is time consuming.

Well, the way Firaxis and Breakaway did all of these was to first make 3D models and animations and then render them out frame by frame to make a 2D animation. This included buildings and most terrain. Firaxis will actually save a bit of time by skipping a step.

 

[Nexushyper]

The major issue with the Civ AI is it's inability to plan ahead or access it's chances for success.

Database AI. Lets look at a game that was poorly.. well, I wont say, but the AI of this game was database driven. SuperPower. The AI had access to the use of statistics and probability. That is the first time you played the AI is was fairly nuetral and some what easy. But the more you played the better the AI became and the harder it became. The database would store the decisions it had made and store it for each nation (in this case civ). And the next game you started it would look at what it did last time and then forecast out a prediction of 5 10 and 20 (not exact but you get the idea) turns ahead.

So say it's Englands turn to move/do something and it wants to move ships to say off the coast of the U.S. and the U.S. is the human player it will look into the databse for England and see if it has done this before. If so it will look and see what happened then and make a prediction based on stats and probability of what will happen this time. It may decide that since last time they got nuked shortly afterwards by the human player that the best thing to do is move ships to Canadas coast. Or it may decide to nuke first. The more you played the better and bigger the database became for the AI to use and thus you had a hard time playing the same way again. Each start was different. If you continually pissed off say Russia every game you played chances are Russia will see this in the database and be more aggresive and more like to attack you before you are ready to attack it.

I thought this system was ingenius! I mean, after so many games of Civ 3 I was able to predict the AI movement and each game I learned and stored in my brain (database) alot of info about the AI and how it reacts to certain things. And thus over time I was able to do things like move 3 units into enemy territory and pull the AI's military in that direction. Then 3 turns later land a dozen transports on the opposite side of the enemies territory and freely walk thru half a dozen cities with only city defenders because the AI sent them after those 3 units... anyways..

AI given access to the ability to forecast and predict and well learn via a database would be a very realistic AI. I hope Civ 4 uses these newer advanced forms of AI for the game.

 

[Ozymandous]

Actually Deep_Blue, the transition from 2D to 3D does not suggest more work on graphics than on the AI-for two very important reasons:

1) Within Firaxis, these different jobs are not a 'zero-sum game'. You have certain members of the team whose primary task is graphics, whilst another part of the team is almost completely devoted to things like the mechanics and AI. These two groups work largely in parallel, though I wouldn't be suprised if there was occasional cross-over.

2) I have it on very good authority that it is in fact easier to do graphics for 3D games than it is for 2D games, and that the graphics component of the game are usually done almost entirely in the last 6-8 months of a game's development cycle-wheras the AI has been consistently 'tweaked' from almost day 1.

Something which gives me hope for the AI in Civ4, though, is that they are apparently building it with an eye on what they have seen going on in multiplayer games done by Firaxis staff and the various playtesters. For example, they are seeing what 'exploits' players are trying to resort to, and designing the AI to effectively counter them. This means not only will we get a better MP game-because it was built from the 'ground up', but that the MP design could lead to a better SP game, as the AI has already been 'taught' to deal with some of the more 'ruthless' tactics of the CivFanatic !

Yours,
Aussie_Lurker.

Hmm, well if it's quicker and easier to develop graphics than AI then why did they wait until only the past 6 months to focus on AI when they have been playing it MP style for 18 months? At least that's what I read on some of the preview type articles on the 'net.

To me that shows that they are more focuses on the MP aspect, which the vast majority will -never- or rarely use, than the SP component, which will be used a great deal. The MP group may be more vocal since they are generally on the 'net more, but the SP group are what usually will buy the game or not based on word of mouth and SP type reviews.

Civ isn't and should never be looked at as DOOM3 where it was 99% designed for MP and a token bit put in for SP. I hope my limited impression of how the development of the game has gone is off the mark, but we'll see.

 

[Raggamuffin]

I tend to not see things in conspiracy vision so I just think the developers are excited about finally getting the MP part right this time.

 

[oldStatesman]

Hmm, well if it's quicker and easier to develop graphics than AI then why did they wait until only the past 6 months to focus on AI when they have been playing it MP style for 18 months? At least that's what I read on some of the preview type articles on the 'net.

To me that shows that they are more focuses on the MP aspect, which the vast majority will -never- or rarely use, than the SP component, which will be used a great deal. The MP group may be more vocal since they are generally on the 'net more, but the SP group are what usually will buy the game or not based on word of mouth and SP type reviews.

Civ isn't and should never be looked at as DOOM3 where it was 99% designed for MP and a token bit put in for SP. I hope my limited impression of how the development of the game has gone is off the mark, but we'll see.

Good post. My fears are the same Ozymandous. You are not alone. I really hope it is the case of overzealous MP designers crowing about thier achievement- but the almost total silence on DETAILS of the SP side has me extrremely concerned. As a buyer truly on the fence, right now the evidence has me jumping down on the side of not getting the game. All things seem to point to a game designed as MP, with SP as an afterthought.

1) Graphics aimed for the younger demographic - who will more likely have time to MP.
2) Complicated combat promotion system that I see an AI having trouble with if not perfectly designed.
3) Use of Great People to Culture Bomb...again, a difficult task for an average AI.

I hope I am wrong. I hope the AI is wonderful, and the SP side is superlative. Firaxis is the team to do this...but the lack of detail is getting very worrisome.