Civ3: Hopes and Aspirations

[Shadowstrike]

The last time I was here, I might have jdged Civ3 a litl too harshly. With luck, Civ3 will not turn out as bad as the more pessimistic posters would make of it. This post basically lists out some of my hopes about Civ3. Feel free to add your own, and start a conversation (yes, I know this is obvious).

One of my hopes has been increased flexibility for scenario editors. Tons of free tech and unit slots, ability to change the graphics around, and so forth. Looking at what the limited resources of Civ2 offered, and what people made of it, we could judge that with increased scenario editing ability, true masters of scenariocrafting can create masterpieces that are to Civ what Picasso was to art or Shakespeare was to prose.

Another hope of mine is that the screenshots that we've seen are not representative of what the final product is like (anyone remember the original SMAC shots?). I would hope for montain ranges to look like mountain ranges, and not small puny individual squares, and deserts to look truely parched. At the same time, cities, units and what not would also look equally brillient.

A third thing I would like to see are new ideas that gradually grow onto the framework of Civ. For example, SMAC introduced borders, artillary bombardment, and so forth. Civ 3 should of nessceity introduce new concepts as such along the framework of its predessesor. Urban spawl, growing city radii, a working resource model, religion, civil policies, and so forth, would be good examples of this (this was off the top of my head, BTW, no need to actually rush off and implement all of them ASAP, though it would be nice)

Finally, like any sequel, I would like to see greater quantities of units, techs, civs, terrains, trade goods and so forth. These new additions should not seem to be plugged into the old framework, but rather an addition that seemed as if it always belonged. Recall how Elephants, Pikemen and Alpine Troops all added themselves so neatly to Civ2. Remember how the reworking of the Civ tech tree was fluid. For optimal effect, the number added shouldn't exceed twice the number of the original, but should come close.

[Wiglaf]

I noticed that you didn't include multiplayer or AI - those definitely should've gotten top honors on your list. Everything else is pretty much dead on: graphics, units, features.

btw aren't you a pessimist?

[Shadowstrike]

Ah yes... AI. Well, i thought that much improved AI was on everyone's lists, but I guess a user-customisable AI would be good. Or an AI which learns from player actions. If you are bulding a super science city, it will lern to do so in future games. Eventually, the AIs will get so good that we'll use them to battle each other.

As for multiplayer, I don't really care either way. Good multiplayer is great, but since I rarely ever play anything multiplayer (and my bnet accounts get hacked on day 3) it wasn't an ipmportant thing to me. However to others, it may be different.

[Simpleton]

(1) Graphics: I thought the graphics looked fine in the screenshots; not "groundbreaking" but fine for me. The only thing I would hope Firaxis would change is the way the roads cross over mountains; it really appears shoddy.

(2) Culture: For me it is probably the most anticipated feature and I really hope Firaxis can make it blend nicely into the game. Since this is one of the most hyped features of the game I think they will ensure it works well. My only concern is with all the new features can they integrate them all seemlessly together? I doubt it but I hope I'm wrong.

(3) I hope the extra CD that comes with the limited edition is worth the additional cash. I would like to see it have indepth video footage and information about the development of Civ3 not just a cheap job thrown in to get people to shell out more cash. Firaxis is a pretty good company though so I trust them.

(4) If we're lucky the editing tools promised with Civ3 will be as good as Firaxis says. I did a little editing in Civ2 but would be interested in doing more if it was easier.

[Wiglaf]

(3) I hope the extra CD that comes with the limited edition is worth the additional cash. I would like to see it have indepth video footage and information about the development of Civ3 not just a cheap job thrown in to get people to shell out more cash. Firaxis is a pretty good company though so I trust them.

Indeed Firaxis is a good company, but their track record isn't perfect. Sid Meier's Antietam was a rehash of Gettysburg (a great game). Yet somehow, Firaxis botched it, making the game far too short and all to full of gimmicks (civil war paintings, a war history section - nothing useable, especially the latter, marred by a horrific interface).

Over the years they've gotten their reputation restored, but don't think they've been spotty clean since their creation.

[Simpleton]

Originally posted by Shadowstrike
Ah yes... AI. Well, i thought that much improved AI was on everyone's lists, but I guess a user-customisable AI would be good. Or an AI which learns from player actions. If you are bulding a super science city, it will lern to do so in future games. Eventually, the AIs will get so good that we'll use them to battle each other.

I was thinking of that same concept. I wonder how hard that would be to develop?

A simple technique might be for the AI to record units built and upgrades made to cities in a text file and then based on the players score decide to save the text file for use in the next game (ie good score use text file. bad score toss it). The AI in the next game you play would then use the saved text file to decide how to proceed. (ie. you got a good score last game by building loads of bombers and getting hoover dam --> next game the computer would race direct it's efforts to discovering techs for bombers and hoover dam --> once it did it would build hoover as fast as possible, build lots of bombers, and since the player liked to build bombers too it would build defences against such an attack like lots of fighters). I don't know how well the AI would use the units or improvements once build but it probably would increase the difficulty for a few reasons. (1) Even if the computer doesn't build the hoover dam project or any wonder in a good spot (large continent) it would still keep the player from getting it. (2) building lots of units to defend against a players favorite unit would force the player to use different techniques (ie computer builds lots of fighters because the player likes bombers --> thus the bombers get shot down. --> therefore players strategy is twarted somewhat) (3) if the AI was actually good at using the units the player liked it could pose pose a problem for the player.

The real difficulty would be to have the AI learn actual strategies from the player (ie how to wage war). The only simple way I thought of that really only partially addresses this problem would be for the AI to make a record of the most popular units a player used (ie. used mostly tanks thus the AI uses tanks too) and/or for the AI to record popular unit combinations the player uses when building armies (ie. player uses the combination of 5 tanks, 3 infantry and 1 bomber quite often so the AI would try this as well).
In the end this type of method could only really be used if the AI could update the saved records it kept as the player changes his/her strategy. This method could also only really be something that the AI used as an aide; it would still have to used it's hard programming for the most part.
I don't know how well this would work but it would be interesting to how well it works.
To any programmers out there: could this idea be implemented well in a civ type game?

[Sabre2th]

I've often thought about having an AI that can learn, but I never thought about how to do it. I always assumed it was beyond our grasp, but maybe not. Unfortunately, we will probably not have a strong, smart AI until true artificial intelligence, which is waaaaay off in the future, if it comes at all. (Though I think it will come sooner than most think)

[Nadexander]

Originally posted by Simpleton
The real difficulty would be to have the AI learn actual strategies from the player (ie how to wage war).
...
To any programmers out there: could this idea be implemented well in a civ type game?

The best thing i think would be to release a preliminary AI and then wait 3 months, browse the apolyton boards, find the dominant strategies that the players use, and then patch the AI to use the same strategies.

-Nadexander

[Simpleton]

Originally posted by Nadexander


The best thing i think would be to release a preliminary AI and then wait 3 months, browse the apolyton boards, find the dominant strategies that the players use, and then patch the AI to use the same strategies.

-Nadexander

That would be a great idea. I think you'd find a lot of happy people if a gaming company did that. It really extends the playability then. I guess the real trick would be to program the AI so it uses the strategies correctly. At least I think that would be the hard part.

[yin26]

I've often thought about having an AI that can learn, but I never thought about how to do it. I always assumed it was beyond our grasp, but maybe not.

Submarine Titans promised and somewhat delivered on that promise. For example, if you came at the AI with one kind of weapon, it would make new craft with the correct counter measure. So, if you played the same way all the time, eventually the comp would just pound you. You had to readjust your strategy to keep the AI files 'guessing' so to speak.

However, unless the AI is programmed properly to handle its economy etc. to begin with, even that kind of adaptive combat AI will have limits. So if you ask me:

** The Civ 3 combat AI should be adaptive across games. You make a profile, and by your 3 or 4th game playing the same way, the AI 'magically' has all the right counters, forcing you to mix things up.

** The part of the AI that REALLY should be opened up relates to the econ/city management stuff. If it's too hard to program an adaptive AI for that, then perhaps finding the "killer build orders" off this and other sites would be the way to go.

** As much as possible, let the fans mess with the AI and make it easy to share AI files with each other.

[Imran Siddiqui]

Sid Meier's Antietam was a rehash of Gettysburg (a great game). Yet somehow, Firaxis botched it, making the game far too short and all to full of gimmicks (civil war paintings, a war history section - nothing useable, especially the latter, marred by a horrific interface).

Ah, but you forget... Antietam was made by BreakAway Games, and had no direct programming from Firaxis. They basically let BreakAway create a game using their engine. Antietam cannot be blamed on them whatsoever.

[Shadowstrike]

Well, my ideas are mstly summed up by yin. Allow "trading" of AI files, manual editing, and an AI that adapts to player tactics. One can only wonder how difficult Xin Yu's AI will become after a year .