Ask The Civ IV Dev Team

On Monday, we asked you for questions for industry legend Sid Meier. Today, we're asking for question to put to the folks behind the technology of Civilization IV. Besides the actual coding and development that went into the game itself, the team has made Civilization IV infinitely moddable through technologies such as XML, Python, and a fully developed SDK. Led by lead designer Soren Johnson, the team will answer your questions about the creation of the fourth chapter in one of the most influential game series out there. So, fire away with your questions. One per comment, please, and keep them topical. We'll pass the ten best questions to Johnson and the team, and the answers will be posted as soon as we have them in our hands.

Comments

The Civ4 AI

[Skyshadow]

My only question for Civ4 concerns the AI: Have you made it a crafty enough opponent yet that it can compete at the higher skill levels of the game without resorting to the "cheating" that we've seen in previous incarnations of the game?

If so, how?

As a player, I almost always find the key to really taking control of a game is to react well to the overall shape of things. Nuances with the terrain, the way cities are arranged in respect to each other, grabbing some resources at the expense of others -- this all provides opportunities for the human player that I wouldn't think an AI could easily pick up on. How can you get the AI to "consider the map", so to speak, rather than simply reacting to the stimulus around it and carrying out a set of predetermined functions (which, at least in my estimation, is the limitation that prevents it from competing fairly at high difficulty levels in the previous Civ games).

Or does the AI find its effectiveness in, say, it's ability to reexamine every city every turn? Or will it, you know, just continue to cheat to compete at advanced levels?

Thanks!

PS: My wife's traveling on business most weekends over the next couple of months. If you wanted to, you know, mail me an advanced copy... Just tossing that out there.

Re:The Civ4 AI

[Silverlancer]

Cheating AIs are always lame. AIs that rely on doing things faster than humans could (see RTSs) are also lame.

A while back I played a game called Galactic Civilizations, a 4X game set in space (compare to Master of Orion). Its best show was the AI, which on the high difficulty levels is simply ingenious. It can spot when you're plotting to do something before you're even half ready to strike, and stop you, without cheating. It is very hard to distract the AI on hard difficulty levels using bait, or any of the classic anti-AI tricks. Even the old tried-and-true get-the-AIs-to-shoot-each-other-until-you-overpowe r-them trick does not work--they will notice that you're circling over the battling AIs like vultures, and team up to kill you.

Re:The Civ4 AI

[freidog]

Just to follow up on that,
Is the AI going to be as moddable and customizable as the rest of the game content?

I know Mr. Caudill mentioned an 'AI SDK' for 'experienced programmers' over on the IGN Civ 4 preview to tailor the AI to their desires. But it was mentioned as a seperate entity from the XML unit files and the basic Phython scripts.
Is this because the AI is more hard coded (less of it in easily accessible scripts) than say unit stats, or just an attempt to give a helping hand to less experianced modderings in a rather complex enviorment like the AI.

Basically I was hoping you could go into some more detail on what AI and other more complex modding might entail.

Re:The Civ4 AI

[SatanicPuppy]

The thing that annoyed me most in CivIII was that it was almost impossible to get a solid technological lead.

You could have a commerce/science race, with all the science buildings built in all your cities, with at least as many cities as the next two largest races combined, and a population that was more than the top FOUR other races combined, with 60% of your GDP going into science, and all the science wonders in one coastal city producing nothing but science, with more science specialists than citizens, and you'd STILL have trouble keeping ahead of other races.

Now, that's just not realistic.

Re:The Civ4 AI

[Keebler71]

Civ 3 is significantly more difficult than Civ 2. In my opinion, the game is much more balanced so previous strategies of focusing on just one thing (war machine, science, etc) simply do not work. These resources are invaluable.

I usually play on Emperor level and probably win 1/10 games although most of the games I know within 10-20 turns if I have any shot at all. Some suggestions:

• Sadly, you need a good start. You simply must start at a good location - preferrably with flood plains or bonus food nearby.
• Expansion. A good fast-growing start allows you to expand quickly (although never as fast as the AI). Once all the land is "claimed" the only way you can expect your boundaries to shift is through miltary or cultural superiority. You need to maximize your area so that...
• Control of resources. Resources. Resources. Resources. They keep your people happy, let you build (or deny the opponents) cool things. Most importantly they are critical for diplomacy
• Keep your opponents happy with you so that their superior militaries don't steamroll you. This is achived through continuous trades of resources and techs. When you do trade a tech, always trade it to EVERYONE the same turn or else the computer will simply turn around and sell it to those you missed. Always trade dead-end techs but never trade techs that lead to important wonders or units (unless you are very near completion of the wonder).
• Avoid alliances as they will drag you into wars. On the other hand, there is nothing like having the AI waste resources fighting itself while you stay neutral and progress in peace. There are ways to encourage the AI to fight. For instance, if you acquire an enemy city that is not very useful to you, offer it in a trade to an opponent of the original owner. They will often start a war to reclaim their city. Oh, and when they do fight, always try to help the underdog. You want this fight to drag out as long as possible. Have settlers ready to move in to the unclaimed territory immediately as one civilization sweeps across another leaving "unclaimed" territory.
• Emphasize cultural production. The first things built in every city should be a temple and library. In my opinion, if you planned your cities very well, then they will grow so fast that they will hit a population limit rather fast in which case the granary isn't doing anything for you. (The exception is slow growing cities) Culture on the other hand is cumulative so it is imperative you build them as early as possible. Having superior culture may be the only way to expand your borders in the mid-game where you likely will still have an insufficient military.
• Prebuild. When a new tech looms on the horizon, start building a dummy improvement early so you can switch to the new wonder/improvement as soon as it is available.
• Use workers to pump up cities. When you run out of expansion room, you may still have some cities that are at max capacity but still have excess food. Use them to make workers which you then assimilate into your smaller cities to make them grow faster.
• Don't try to build every wonder or acquire every tech first. Be tactical. Make a break for philosophy first. If you get it first, you get a free tech. Use it to get to Republic first and change your government. Culture, growth and production will explode! Try to build the Great Library which will keep you on par with your peers in science through the early mid-game while you investigate paths they are ignoring. Make sure you get railroads and are ready with dozens of waiting workers to start building immediately. Switch to democracy at first opportunity and hold off on communism/fascism untill the end game when corruption is out of control.
• Save and re-try. I know it is cheesy but the best way to learn from your mistakes is to play though several different possibilites and find what works best for you.

You will always lag behind the AI's technology in the beginning, but by the mid-game you should be pretty even and in good shape for a late game explosion.

Re:The Civ4 AI

[cdsr]

Let me add a few (not in response to GP, as a disclaimer: I can beat Diety pretty frequently).

- On low difficulties build order should be 1. warrior, 2. settler

- On high difficulties build order should be 1. warrior, 2. granary (if you don't have pottery research it first, build a dummy improvement then switch), 3. settler, 4. settler, etc. Keep pumping settlers from here until you run out of room. If your starting city doesn't have access to cattle/wheat then find a city that does and make that your settler pump. More than one settler pump is good if the map is big. You will soon find that you've caught up to enemy civs in terms of number of cities.

- Those settler pumps should mix in workers as well along with cities that hit their population limits (i.e. they are wasting food) ... connect those cities! and get all resources and luxuries ASAP.

- Resources are worth going to war for. Some more than others but Rubber is king, see below.

- If you don't have the resources to build tanks/mech inf/modern armour you can still go on the offensive against those that do (hopefully to obtain the resources you need). Build enough artillery to be able to do the following in a single turn: 1. reduce a city to population one, 2. destroy all improvements, 3. bring all units down to one hit point. Then use one or two armies (the unit, made from leaders gained in combat) of cavalry to take out the last hit point. Use infantry for defense.

- Artillery, if used correctly, is devastating.

- Create a bigger industrial core by arranging your cities in a ring around your capital, the game calculates corruption partly by the number of cities in between the target city and your capital. Use the Forbidden Palace to extend this ring and create a large industrial core. Cities very far from your capital but close to a FP aren't that great anyway. The overlap turns those mediocre cities near your capital into powerhouses.

- Building cities with two spaces in between (CxxCxxC) is useful for two reasons: 1. slow units can jump from city to city in a single turn using roads, 2. extra culture isn't required to connect city boundaries. But there will be some overlap in tiles so the cities won't be as big.

- At high difficulties there often isn't time for both a temple and library. Build libraries for the science bonus. Temples are only good if you're a religious civ.

- Specialists can turn large but completely corrupt cities into industrial powerhouses. When available, turn the excess entertainers into civil engineers. Each CE can give two shields per turn towards a city improvement. Use these for culture and science. Other large cities with excess entertainers can use CE/police/scientists as well. I jumped a complete difficutly setting when I started using specialists properly, they give huge bonuses.

Portables

[BMonger]

Is there any chance we'll get to see some of the Civ titles moved to portables? I think the game would play wonderfully on the DS.

My Question for You...

[BishonenAngstMagnet]

Will the game support Internet as well as LAN play? How extensive is the multiplayer going to be (if any)?

Addiction

[MBraynard]

As you are probably aware, your games are highly addictive. They have inevitably lead to some screwed up lives - failing classes, getting expelled, losing jobs, losing significant others, losing families, losing a lot.

By making such a good game, do you think you are culpable of the effects of the game _at all_ and are you thinking about putting in counter measures to allow people to better set limits for themselves within the game?

Unit Moddablity?

[EngineeringMarvel]

Will there be any limitations on the moddablity of the units in the game? This ranges from the textures (the way they look), the abilities (can new ones be added), and stats (A/D/B). Or to rephrase the question, what do you expect modders will look forward to the most when it comes to modding the units in the game?

Extensibility

[Frac]

With Civilization IV being so flexible and moddable, do you see a Civilization V in the future? Or do you see IV becoming a platform where new content become expansion pack, just like The Sims franchise?

Family Gaming

[carambola5]

Growing up, playing games with the family meant getting out classic boardgames like Monopoly, Risk, etc. The Civilization games seem like a prime candidate for breaking into the family-game-playing field. What, if any, steps has your team taken to bring your game(s) to the level of "game night with the kids?" What technologies, such as display and control, need to be developed before such an environment is realized?

Politics...

[MosesJones]

How much will CiV4 use political shifts in countries as a cyclical change in approaches (e.g. new democratic leader with a different political alignment will form different alliances).

How much is Sid involved?

[FortKnox]

It seems like every sequel that comes out Sid is less and less involved in the product. How much is Sid involved in 4? Does he help code? Help design? Help produce? Or stamp his name on the finished product?

No bad vibes, meant to the Gaming God... just curious how involved he is with the 5th (counting "Alpha Centauri") cantation of his classic...

How customisable?

[m50d]

Civ3 was wonderfully customisable as long as you were sticking with a civ-type game, but even basic reaching beyond this ran into trouble, e.g. I found no way to get the map generator to have different weighting of tile types, or extend the number of varieties beyond the land/sea split. Have all these kind of limitations been removed? How possible is a total conversion? What about conversions to a different game type?

Civ Economy

[reynard_ze_fox]

Do you have any plans to include other means of warfare, such as economic warfare and the use of interest rates, bonds, companies and good old fashioned money to conquer? It could add another layer of depth once players approach the modern age.

Imagine a small country becoming a trading / banking power, sort of like the Dutch (minus the whole tulip fiasco), or Switzerland, countries that can buy their immunity and economically dominate other countries.

Just a thought...

Do you think 3D graphics will enhance gameplay?

[Anubis333]

As a long time Civ player, I would have to say that I really didn't understand why it moved to 3D graphics. The 3D rendered sprites weren't really comparable to the 2d artwork, and it didn't really feel like a needed addition. Will having the engine be entirely 3D actually add to the gameplay in any way, other than have objects occlude one another?

When I say 'add to the gameplay' I mean, add to the game experience in a way 2D sprites couldn't. For example: Physics, multipls views, wind, etc.. (I have only really seen the 3D globe, and like the idea)

As a 3D game developer, I have seen so many of my favorite games rehashed into 3D versions just because the developers thought that a 2D sprite-based game cannot make it in this market, and that annoys me. From Pirates! to Monkey Island, it seems developers would rather make a 3D game without any real need for 3D art or gameplay elements. Do you feel this pressure, or do you feel that a 3D game is inherently better because it has a new dimention? (Even if it still has the same locked off camera angle and usually poorer quality art assets)

what was the worst technology decision you made?

[Surt]

And what would you do differently if you could go back and reverse that decision?

Working with Sid

[Avacar]

As the Dev Team, how do you feel working with Sid? Did you play his games before you became a game developer? If so, do you feel intimidated working with an icon of the industry? Were you part of Sid's original team? If so, how has project management changed throughout the years on his projects?

civ2 - civ4

[N3wsByt3]

As many, I started with civ1, which was a nice game, for its time. I simply adored civ2, which I have played over and over again, and it continues - even today - to be a game I (re)play now and then. This may seem as no big deal, until one realises I have *never* felt an urge to repeatedly play a game several times; mostly I play it once through, and I'm rather bored by it, be it an RPG, a first-person-shooter, or a strategic oriented game - which usually I like the most and on average I play 3-4 times. After a while, *no* game can hold my interest I've noticed, the notable exeption being civ2.

Alas...when civ3 came out, it didn't do it for me. Despite the poor graphics compared to civ3, I still prefer civ2. It's not easy to put the finger on the the reason why, but I suspect it's because civ3 has become a bit *too* complex. It's all very nice to have borders of influence, and insurgents in cities, and elaborate negotiation...but, somehow, I find civ2 is just easier and more fun to play. Sometimes, one just wants to 'go for it', without all the extra complexity. Now, will it be possible to play Civ4 in a 'easy' mode, which makes it more simple and user-friendly according to the lines (and rules) of civ2? I really think such a 'easy' setting would be greatly appreciated by those who want less complexity, and more simple, user-friendly gameplay.
Alternatively, will you place the civ2 game (and engine) under the GPL or similar licence, so people might freely hack and expand on that?

Classic modes

[AvitarX]

With the extreme modability will we be able to get classic modes of play?

for example will I be able to play Civ 1,2 or 3, and not just their rules, but their units, tech trees and civilipedia?

Will this be provided or will it (if possible) have to be user add-ons?

If they are user add-ons will the team help a serious community effort to help them get the propper algorythems for combat resolution and what not (so our precious bomber can still be killed by the phalax that walks away undamaged)?

Is this one question? I think it counts as such.

Macro and Micro Management

[kenp2002]

How, in constrast to how Masters of Orion 3, will the Civ team be addressing macro and micro management aspects of the game? RTS games are forced to place heavy consideration into managing in real time units and control and the scope of an RTS prevents a snowball effect. Turned based games become burdened by logistical considerations as a result of not having that same focus on micromanagment. Managing 55 workers in Civ3 along with 35 cities becomes a logistical nightmare when governor AI doesn't learn from your play style. Directing 22 to build 13 fortresses across a continent while running rail lines to each with production queues rallied to those location but only to a max of 25 units per fortress ares and having to manually intercept an invading force resultsed in a single turn while playing Civ3 that took 2 weeks to process (thats 2 weeks of play time. It actually took about 2 months to move to the next turn.) Additionally having a stack of 75 units attack a city is a rather dull event, even worse when the computer attacks.

Ken's Rule of Gaming: Complexity in feature should be inversly proportional to the amount of player control.

The more complex a process is in real life, the less direct control a person has, this is what MOO3 tried to resolve.

MOO3 was a real shock to many players but once you learned to let go of micromanagment the game becomes rather plesant and suprising. A good contrast is what Sim City is To Civilization as Civilization is to M003.

Which Direction is Civ4 taking?

Re:Linux Support

[Trelane]

es, Linux is great and all, and we all want to see it become the next desktop platform, but is it too much to ask for you to either dual boot Windows or have a separate box for your Wintendo?

If you want me to buy your game, then yes, it is too much to ask. It's called "voting with your wallet," and I try to practice it. If you want Linux to be well supported, you have to help out. It ain't gonna happen on its own.

Additionally, cross-platform game development needn't add too much additional labour if you start off designing it as such, and generally the quality of the code is better (because you have to sit down and think about how to do things intelligently, not to mention bug fixing).

Now, if you think that Linux support is intractable because of support costs, then fine. Give me the game without support. Just give me some version of the game. Of course I won't pay the same amount as with support, and if someone else does more, I'll be more loyal to them. But it'll eliminate the need for support costs.

Re:Mac Version

[Brad Oliver]

The Mac version of Civ3 left so much to be desired (in terms of AI) in comparison to the Windows version that I finally just stopped playing it.

Disclaimer: I'm the programmer who worked on the Mac port of Civ3.

The AI code in the Mac and PC versions is identical. I've never heard of any AI variances between the 2, and no differences have been pointed out by the Mac fans over at the CivFanatics forums (some of them are cross-platform users). Can you provide a saved game that demonstrates a genuine difference?

Map reactive is cool, but player reactive better

[KingSkippus]

I like the idea of an AI reacting to the subtleties of the map.

But what I would like to see is an AI that reacts to how its opponent is playing.

For example, I would like an AI that evaluates how aggressive a player is. If a player is constantly attacking the AI, it should react by foregoing some research to build up a defensive army. If a player is defensive, on the other hand, the AI should patiently build up an overwhelming attack force, complete with research upgrades and such.

If a player shows a preference for attacking with one particular type of unit, the AI should "realize" it and start building counterunits. It would even be nice if the AI would do things like sacrifice some scouts to find out what its opponents are up to and compensate for it. Does the scout see some lots of planes in a city? Build some anti-aircraft missile batteries in surrounding areas. I've beaten lots of various players at various strategy games using these kinds of tactics. If an AI used them too, it would add a whole new dimension to player-vs.-computer strategy games.

Plus, it would be nice for developers to observe some really good players playing, make some notes, and ask the players why they do things that the developers don't understand. Are there any general rules that can be programmed that a human uses in making decisions like when he or she starts building military units, how those units are deployed, how much and what kind of research he or she conducts and when, and so on.

I think a cool AI feature of a game would be for the AI to "learn" how a player plays over the course of the player's games. If I beat the computer one way, it will know where it went wrong and play the next game differently, under the assumption that the player will still use some of the same tactics. Perhaps a game would even include some sort of profile manager so that if my brother plays the game, the computer will play against him differently. I've used that tactic several times in AoE2—record the games so I can go back later and study why my opponent did to spank me so badly. Next game I play against that opponent, he or she will be pwned by someone who has prepared for his or her tricks and strategies.

I think it would also be cool for the AI to try a few odd tactics now and then to see how a player reacts. Start building a wonder. What did the player do? Immediately start one of his or her own? Use that knowledge to make him or her waste resources that could otherwise be alloted to the military. Declare war on a player out of the blue and see what happens. Does the player start making concessions to re-establish peace? If so, that player can be bluffed. Send a lone military unit to camp close to another city. Does the player attack him immediately, though he's no threat? If so, do the same thing, but have a larger army waiting on the other side of the city to go in while it's not as heavily defended.

I guess what I'm saying is that if we could get to the point where computers are "thinking" like humans, I can finally shed the last vestiges of my need for friends to play with, and that can't be a bad thing, right? :-)

Alpha Centauri

[squiggleslash]

With the dust being blown off the Civilization for the third time, I'm wondering if similar plans are afoot to work on Alpha Centauri, and if so, how the original will be improved upon.

Some of us see AC as the best in the whole (greater) Civilization series. Awesome game.

DRM in Civ IV

[Lord Ender]

Civ III requires the installation CD be inserted every time you play, even though none of the content on the CD is used by the game after installation. This annoys your customers by making them juggle CDs, unnecessarily wear out their hardware, and shorten their battery life. Consequently, many of your customers install "No-CD Cracks" to fix this flaw in your software.
How do you feel about the existence and use of such cracks? Will you include this CD requirement in Civ IV even though it does not prevent copyright infringement but still inconveniences your customers?

A different AI question: AI moddability

[Dr. Spork]

My question is AI related as well. Since the parent is a good question, I'd like to tack this on:

Will the code for the AI routines be user-editable, easy to mod and documented?

Rationalle: As an fan AI-coder for CRPG's (I worked with David Gaider on AI in the Ascention mod of Baldur's Gate 2), it's my experience that with no deadlines and lot of playing experience (very important), a community of modders are willing and able to write a much smarter AI than any game engineer. Nothing would more increase my willingness to replay the game than the promise that this time, a newly modded AI really will give me a run for my money. In my experience, it only took several solid weeks of playing and a few weeks of coding before I could make a computer-controlled magic-user in BG2 who could regularly kick the ass of an identically-able human controlled magic user, without cheating.

For Civ-specific AI issues, here are the features of what I take to be the holy grail of AI:

1. No omniscience: The input information available to each country's AI would be the same as what would be available to a player if she controlled that country. (No "seeing past your range of view".) 2. The AI is completely blind as to which rivals are human and which are AI. 3. There are several very good AI's that each favor different strategies, and a meta-AI that determines which strategy is the best fit for the situation. 4. Exactly the same rules apply to the AI civs as to the human-controlled civs (regarding science, production, trade, etc.).

Globe?

[atomray]

I have enjoyed the Civilization series since the beginning, but with the third incarnation it was pointed out to me that it's rather disappointing that the game continues to be played on a tube rather than a globe. If Civilization IV is also on a strip, could you explain what difficulties the development team is having in implementing what seems to be such an obvious and simple detail?

Python+XML vs lua

[SumDog]

I've noticed many other games have engines based on lua. I believe the first two Warcraft series use lua extensively for level development and is what people wrote custom mods with if they didn't want to use the built in map editors.

Don't get me wrong, I'm not a huge fan of lua or anything. I've only done minor programming in it. My question is why did you choose the language that you did (python + xml files), what are the advantages to this approach, what are the disadvantages and finally, how much development time would you say is needed using your SDK would take vs attempting to design a mod for some of the other popular games (Quake3, Half-Life2, etc.)

Oh and I guess one more thing. How far have we come in modding games since Doom I .wad files?

Sumit

UI design for multiple monitors

[natebrau]

The old Civ II (on the Mac port, at least) was fantastic in its support for multiple monitors. Everything was implemented as its own individual window/palette- the main game screen was one window, the tool palette was another window, animations popped up in another window. This was spectacular for multiple monitors, since the main monitor was free to show the main game screen and only the main game screen, while all secondary activity could be displayed on the second monitor. Civ 3 destroyed this, and brought everything back into one single monolithic ueber-window, where any action brought up a dialog box/window which was drawn on top of the main game screen, obscuring game information. Will Civ 4 continue this approach and assume that everyone must have a single, solitary monitor, or will it go back to a floating palette approach, where those of us with multiple monitors can really take advantage of them? Thanks in advance for your time, -Natebrau