Try *for once* to play the game in a different way. I think there is about 95% chance that you are mad at the AI because you can win so easy in a war.

Well, consider this for a while. Should you ever look in your manual, you will notice there are several ways of winning in Call to Power. With this post I hope to focus you on a different way of winning.

I am talking about the Diplomatic Victory. This requires you to become friends with every nation on the planet. Okay, this immediatly shows another problem. I bet you have already managed to win this way. I also bet you have destroyed many an empire before getting this victory.

Now, here comes my point. The AI is weak in battle. But there is also a way of winning without battle. So try to keep them apart. Go soft, become an UN kind of nation. Lots of talk and declaring war, but no action. Do not build an army, and above all, do not move it into enemy territory.

The strongest side of the AI lies in diplomatic relationships. Again, many people disagree, and say that the AI is bad in diplomacy. Well, it is, in certain kinds of diplomacy. Notably, the war like diplomacy (helping you, withdrawing, asking to stop war against someone etc. etc.)

What many people do not notice is the position the AI takes in diplomacy. From the start of the game, their opinion about you lies lower then neutral. They do not like you. Now, take in account that the only way of winning is by making friends, and not by attacking people. What you have to do, is make friends. Well, believe me, that will take some time. It is actually quite a challenge to get everyone on your side. But it is possible. Also, when your military is weaker, their military will become relativly stronger. So you might indeed lose a city to invaders sometimes, and have a struggle to reclaim it. You will not have such a struggle when you use 6 units to defend each base. Limit yourself to a maximum of two defending units.

To make things more difficult, you can set your own goals. Create the civilasation *YOU* would like to life in. Browse through the governments, and look for the ones you agree with, and not for the ones that are the strongest.

Try to find the strenghts that lie in a game, not the weaknesses. If you find a strategy that the AI just cant counter, abandon it, and try something else.

Also, I would like to tell you something. How hard do you think it is to create an AI that can play a civ-game? You all talk like it is something simple. Well, take my word on it, it is not. I have studied Artificial Intelligence, and create them for a living. There are many things that seem simple enough, but are really hard to create. Just take a simple game, four in a row. When you look at the game, it is easy for you to decide what move to make. But for a computer that is already difficult. A computer cannot 'see' the entire situation. All it can do is 'look at a square', then see what is on it, and perhaps look around the square, to see what is there.

In a shooter-game, all the computer has to do is follow certain path's, and when it sees an enemy, aim + fire. That is easy to create. It is safe to say the hardest part with creating such AI's, is to make sure that they dont outclass the player. They do not have to think. Once they see you, the can kill you with 100% accuracy. Remember, whenever you play UT or Quake or Half-Life: the AI is just letting you win. It is a hundred times stronger then you are. Sorry...

In a civ-game, it is different. A civ-game is about logic. In a shooter it is about reflexes. The greatest part of the game you are thinking about what move to make next. Logic is simple for you, but not for a computer. An AI simply cant see the 'big picture'. That is not entirely true, it is possible of advanced reasonong, but then all the AI's would each take the same time as you do to make one turn. And people dont like that. So Activision had to things another way. This AI just looks at avereges and such, and then decides by pre-programmed tables what to do. And yes, those tables are easy to remember, and once you do, you can easily defeat the AI. But you dont want to wait 15 minutes for one nation to make a move, then wait another 6*15 minutes for the *other* AI nations to make *their* moves??? That is what would be the result of a too strong but slow AI.

In conclusion: it is almost impossible to create a fast AI that is capable of playing excellent Civ. It would simply require to much speed from your computer. So try to find the strenghts of the AI. These strenghts lie in diplomacy. In diplomacy, they do not have to check an entire world-map. All they have to do is pick their anwsers from a list, and give you a hard time to get that allience. And they are good at that...

Please, try to overcome your anger, and play the game in a different way...

Alisia

----------------------------------------------------

In conclusion: a 'short' description of the way I play the game. Read it if you wany, but it is not a real part of my rant...

what u have not understood is that we want to play with someone in front of us: the AI never attacks!
see: ../../Forum35/HTML/000852.html?14
and ../../Forum35/HTML/000871.html?3
and u'll understand our point of vu

i am like u, i never win my game with the bloodlust victory, since civ1 i played all civs an i prefer winning with the alpha centaury, gaian and others alien victory BUT we need a challenge!
why when i am at war with a nation it does nothing to me and let me take all her cities? why in impossibl mode the game is so easy?
when i remember the difficulty i had in civ 1 to win an impossible game(we had to change an octet in the save game from 6 to 7 héhé, t'was the good time) i am asking myself if the AI developpers have'nt take vacation for years??

after that u talk about the difficulty to make a good AI but the Ai of civ 1 and civ2 where better on that point: they attacked me and i played it on a 80286 machine with 640ko of memory so dont say me it require important cpu to make a GOOD AI
that AI is bad and bad developped sorry, watch the other civ and other game, a AI which never attack isen't challenging

another thing: we have seen on this forum the thing to changes in the files of ctp2 to increase the intelligence of the AI! why the activision developper haven't tested enough the game to see that the AI never attacked and change themself these settiong before selling the game?

the AI might be good if the settings on the txt files are good, that's my point of vue

i pray for a 1.2 patch that could correct this

If I wanted to play a game based totally on diplomacy, well, I would have bought "Diplomacy". Instead I bought a game that is supposed to take different elements and blend them into a cohesive product with different outcomes possible. A very large part of the game is a war-based game. There are more military units than anything else in the game. If they are part of the game, they should be used. The AI doesn't... it has in the past.

I've been playing the civ games since Civ-1 on my 286, just as Bed23 has. I'm not a rookie to the genre and can appreciate what has to be done to create a challenging product. In all the Civ games, I've been attacked offensively, sometimes even with great fury, by the AI. On less 'advanced' machines, in more or less the same games decent offenses have been formed. Not a lot behind the scenes has changed from Civ1/Civ2 to SMAC to CTP2 besides graphics/interface/names of units/names of technologies. Yes, they've become a little more advanced, but not as advanced as the technology has gotten. There is no excuse for the AI to sit there like a dead fish.

The standard out of the box game stinks rushed product long way. A game as complicated as this one need to be playtested and balanced/tweaked for maybe a year atleast. Activision simple rushed it to get the money. No gameplay quality assurance. The standard is fun the first few times you play it, but a game like this should be fun more than the first few times. Compare it to CIV 2. The AI in CIV2 atleast try to attack you. Which makes the games you play more fun and replay value high.

I have not given up hope though, for with some tweaking and a patch fixing some stuff with diplomacy and AIs attacking alot more and fixing the issues with Air units /Sea units. Until then I Play Europa Universallis great strategy game and great support.

One more thing BIG credit to the mod makers of CTP2 without them there is no hope. :-) The team of CTP2 needs credit aswell they are not to blame they got rushed by the money hungry executives.

I am just a sad Civer that see the potential of this rushed game get wasted because money is king.
The game creators had legendary status in their hands and they had to let it slip through their fingers.

Hope SID can do CIV 3 right out of the box.

I understand what Alisia is saying, but still she must admit theres something unfun about a strategy game when you can steamroll over your opponents so easily.

I'm a little put off by this. I haven't played for a week or so. In my current game I'm at about 2270 and I'm building lots of monoliths and satellites and stuff. Rather haphazardly because the exact requirements for that kind of victory weren't in the manual. The only way activision can get back into the lowest rankings of my good graces is to at least come up with a patch to make the AI more aggressive. I've attacked civs in this game out of sheer boredom.

But aren't you forgetting something? Sure, it's good to have several ways of winning. Sure, it is good to have some kind of diversity to make all the games different and challenging. But us boys, well, we like a good war. And to be frank, diplomacy is for girls. Kidding.

Actually, my point is that although there are several different ways to win, winning through war is one of the most fun and should be viable. Activision didn't do a good job with the AI. The fact that war is not fun because the AI is a wimp is a sign of some sloppy design. People are right to complain because a crucial element of the game doesn't work as well as it should. If you look through history, war has played a very important part in the establishing and downfall of civilizations.

Having said that, its no fun in a game like this to have in the back of my mind, the thought : "Well, if things start to go south, at least I can cream them militarily."

On, to quote popular culture, "Yes, but droids don't pull your arm off when they lose."

YouknowwhatImean?

Leave it up to a female to want to play some woosy "can't we all get along" diplomatic version of CTP2.

As far as the changes that others have made. Most suggestions have explanations behind them as to why that person thinks that change would have an effect. Thats not a random change, but a thought out change.

There's nothing wrong with playing any of the possible winning strategies. But when you have NO military threat at all, it gives the human a large advantage because they are concentrating on science and diplomacy whereas the AI supposedly is playing a more balanced game. Before I started making changes, my interior cities would be guarded by a single warrior (if any at all) and most of my outer cities contained a warrior and an archer. I'd have one or 2 stacks out attacking. My military expediture was around 5% whereas the AI civs are around 20% or more. So even playing a non military strategy, the human's economy has a large advantage of not supporting an army to defend against the large but useless AI armies.

Try making some of these 'random' changes and see how it affects your play when you must play an active defense.

BTW Alisia, I believe I've come up with a working model of an AI that will think on a tri-level basis using simple yes/no comparisons. EG. Does player X have more attack value than me? Is this area more defensible than this area? And so on. My model can come up with three sets of objectives. Level 3 objectives are controlled by the city mayors and based on orders from objective 2. Objective 2 is split into the five national areas of military, economics, science, domestic and foreign. These objectives determine national-level indicators and non-city troop placement. Objective 2 is influenced by objective 1 and influences objective 3. Objective 1 is the overall "AI overlord's" stance for the entire nation. For instance, using simple yes/no questions, the AI overlord can determine that nation X is weaker militarily, economically and productive wise. Therefore it determines it can take it over. Objective 1: take over nation X. The five objective 2 areas look at this order and with simple yes/no questions can determine that more attack units are needed, more units on that flank are needed, that trade routes with other nations are needed to compensate loss of nation X's routes, and foreign needs to declare war. Individual mayors look at their cities using all the objective 2's and determine that cities in the war border area build attack units (based on a simply distance_from_border rule), and other cities build caravans. I'm in the process of drawing all this into a Visio flowchart, and will set one up in C++ to see how it goes. I might just simply feed CTP2's values into it and see what it gives as Obj 1, 2 & 3.

quote: Originally posted by Alpha Wolf on 01-04-2001 05:27 PM As a result, I have improved AI performance to the point that I've lowered the starting advantages for the AI on impossible and its 1000ad and 2 civs are ahead of me both militarily and technically. their defenses have reason behind them, and have effectively held off many of my armies. The barbarians now have some logic behind their madness and will actually produce things when they capture cities. Both AI civs and the barbarians will stack when threatened and unstack otherwise. And both have launched successful offenses against not only each other but also against me.

Alphawolf,

I can understand both of the major point of views in this thread - being a nasty girlie :-) I would like both - ie to be free in the decision which way I want to play. I keep on reading about AI changes in the Textfiles, however I am pretty confused about all of the changes which are proposed by a lot of people. I would like to start playing a game with the AI modified in your way, so could you please let me know what exactly you did ? Is there any way you could either make the Textfiles you changed public or send it to me ?

Thanks a lot for reading this,
Martina

Oh, and that bluevoss@bellsouth.net, if you're sending any out...

I've heard that the science victory in CtP2 (haven't attempted it) is the most difficult ever for a civ game.

Looks like my site accessible now. Have fun beating back the barbarians

I like the diplomatic victory but diplomacy is also based on military strength. With a nice big military its easier to bully the AI into agreements. In the science victory you need 60% or more obelisk coverage. If the AI was aggressive it would see your strategy and unalliance from you and start pilliaging obelisks to stop you from winning so you should need military coverage to ensure a win. I played one game where the AI did in fact start pilliaging all my obelisks or at least tried. I have yet to play a game so far where the computer controlled nation starts or completes the science victory.

I personally dislike the diplomacy victory. I find it too easy to bully the AI into alliances. I also don't like the fact that the computers rarely if ever form them between themselves and try and beat you to the alliance victory. Has anyone yet lost to an alliance victory in a single player game? I hope not. Forming that last alliance the AI would need is equivilant to hitting the power switch. I like the concept of global alliances but it just doesnt seem right. Why would any player AI or human agree to an alliance knowing full well it ends the game with a victory for someone else? Besides, there is always some nation out there that won't play nice with everyone else.

Since the computer doesn't seem to like to try to beat you with the diplomatic victory, and it doesn't seem to o rushed to beat you with the science victory the only thing left to fear is a military defeat. And now we are back to square one, it doesn't even try to beat you that way very well.

Does the AI consider who else you allied with when agreeing to an alliance?

If AI1 and AI2 hate each other and you are allied with AI1, wouldn't AI2 be VERY VERY reluctant to sign an alliance with you since you are already allied with a mortal enemy? Or perhaps by signing an alliance with AI2 it should automatically cancel an alliance between yourself an AI1.

Seems to me that a true alliance victory should require global peace and alliances between all nations with each other instead of just one common one.

Ive been programming for the last 15years, and AI coding for the last 5 years in various applications .. be it a simple lisp knowledge/action tree or a complicated second guessing AI of the neural nature in C .. The only reason I can see for an AI that is so lame in attack .. is if it was coded to learn .. and learn off you .. the only problem is, we all know that CTP2 doesn't learn anything.

This idea that somehow you need some amazing super computer to deal with complex AI is crazy .. yes its true, the more complex the game, the more information is needed to store, but if you use neural nets or even basic tree structures you can bypass the 99.9% of irrelavent info, and get straight to the answer when requesting an action.. it only require more memory, or more HD space .. not a faster CPU due to the nature of the search algorithm. The other issue is this, all we want is an aggressive foe, if the AI decided to build as many offensive units (and use them) as it does defensive .. we'd have a game .. that doesn't even require a complex AI .. just a change in rules governing how units are instructed and built.

quote: Originally posted by The Viceroy on 01-05-2001 08:32 AM ...Ive been programming for the last 15years...

Really?

-just kidding - I've been complaining about CTP2 for (what feels like) 15 years

I remember when I worked freelance for SSI years ago. A guy there was making a space game where you would set up your power and shields and course, esentually try to out-predict the computer. Well, I asked him how difficult it was to tie AI into a 0-100% AI selectable level. He said not hard - the skill level was the percent chance that the computer would "peek" at your move and plot the best way to nail you.

Which makes we wonder on something like CTP2 - I wonder why they don't build a number of (for lack of a better word) attack runs. You know, make a map with invasion routes already established, along with potential city sites, and the like. That way, the computer could break things up into regions, and handle it accoridingly. Its sorta like building the world around the stratagies, rather than the other way round.

Of course, about the third time you saw the computer land on the same beach, you would grow suspicious. And it would make building worlds a lot harder.

Just thinking out loud.

At this point, it seems the AI is tweaked to my liking altho I've only played into the early 1000s AD. But I'm assuming that it'll still follow the basic rules i changed even as it gets better units.

Things I'm still working on are:
Generals;
Unit balancing, replacing game values with more realistic ones;
Natural Disasters,
Balanced AI stacks (5 front line, 5 ranged, 2 flanker);
Barbarian specific units;
AI naval strategy, not exactly sure why its so different than land;
Use of terrorists by the barbarians late in the game;
A mod builder that will have use friendly interfaces to make changes, then those changes will be propogated throughout all the necessary files. It will also allow different mods to be compared on a file basis. I've mapped most of this out, but time has become very limited lately so this will be awhile in coming.

With few exceptions, I'm using the advances/buildings/units that came with the game. Much later I may play around with a new tech tree. Having a fairly good idea of how the AI acts, lets me know how to set up a tree that even the AI will benefit from. Too many moders just add things that the human will use, like we need more advantages.

quote: Originally posted by Alpha Wolf on 01-04-2001 07:58 PM I'll put my "website" on my apolyton profile so people can d/l at their own risk.

Alphawolf,

Thanks a lot !!! I will play with your files on Sunday and will tell you how I feel about them !

quote: Originally posted by Alpha Wolf on 01-04-2001 07:58 PM I'll put my "website" on my apolyton profile so people can d/l at their own risk.

Alphawolf,

Thanks a lot !!! I will play with your files on Sunday and will tell you how I feel about them !

We know the military route is flawed as it is too easy and the AI is unresponsive and "witless". Several folks have recommended the diplomacy and scientific avenues of victory as alternatives. Well, I disagree. Even if there were no such thing as individual differences in preferences for gameplay like military, science, diplo, the argument in favor of using one of the two alternatives to military strategy is still flawed.

Why do I say that? Because, the diplomacy AI model is as witless and unresponsive as the military AI. I am one of those who really would prefer a diplomatic strategy to victory but the AI is toothless or "zoned" almost always. I offer cities for a ceasefire when I am the one rampaging through it's empire and it says "no"? I offer technology, even military tech for a map that is only covering half of what I already have discovered and it says "no"? This is no one-time thing. This goes on and on. So this route is dead and I take up arms and cream the opponent civs. Whoopdee.

OK, so let's race the civs to some superior technology level! No...let's not. The tech tree is boring and wrong and twisted and I can't buy into the "alien" thing. I used to be an X-files fan too, yeah, six years ago! Got the X-file "boot-up" theme on my PC back in 1994 and it was so cool. So, well... the only X-files thing I have any interest in now is filling my shiny midnight black X-files coffee mug with French Roast. Thank goodness no picture of Molder or Scully haunt the elegance of the cup's appearance or it would have gone out in the last garage sale.

So I too am among the CTP diaspora.. lurking, waiting for a fix.. not having played in a week and very interested in that "other game" over in the "other game" forum.

Value might be the value of the attacked target
power might be the power of your empire.

Just guessing, thats the most frustrating about editing this game to much guessing.

Anyhow, firstly the point of game AI which it seems you have missed in both Civ games and shooter games is not to let the computer play at "its best" but to simulate the different ways a human can play - allowing us mere mortals without friends or net connections a realistic game. Hence when we play civ we expect some of the opposition to attack, some to be defensive etc.. (oh and in FPS games some bots need to be good some to be crap - I'm sure the people who created the AI in FPS would not appreciate the of the cuff comments you make - I know the work they do is not easy! (How do I know? - I studied AI (finished a few years ago thankfully!) and often review games for a radio station so sometimes speak to the dev people.

You then go on to say "play the game in this way" - well the game was designed to allow us to play in this way and many others - if the AI could not cope (and it doesn't) then the game should not have been released till it could.

As for saying that "it is almost impossible to create a fast AI that is capable of playing excellent Civ" - yes true with some forms of AI, but there are many different forms of AI some simple some complex and games have been making use of both for a long time. AI has moved forward rapidly in recent times. CTP II is not majorly complicated compared to Civ 2 and certainly not very much more complicated than CTP - yes there is more diplomacy options but if they couldn't implement it properly surely it was better to wait till it can. That way they dont antagonise their customers who of course if annoyed enough will no longer purchase the next CTP - surely not good for the long term futre of CTP?!

My guess is CTP II AI can be fixed and is being part proven by some of the mods out there and hopefully along with the rest of the people out there the next patch will fix it!

quote: Originally posted by WesW on 01-06-2001 06:52 AM Hi everyone. I hope you guys can help me figure a few things out here. In the goals.txt, there are lists of values which the AI takes into account to decide what to do. Well, there are three that I can't figure out. They are: ThreatBonus, ValueBonus and PowerBonus. The goals page of the modification section has a "very" brief definition of these terms, too brief for me to be able to tell what values they read from and what their affect is. If you could help me out here, especially with the difference between Value and Power, then I might be able to figure out just how the AI makes its decisions. Value is a term referenced in Strategies.txt as one of the things used to determine whether to attack or not, but how is "value" determined? Power is a line in units.txt, a value of which is assigned to each unit, seemingly close to the construction cost of the unit, and used for computing the power rankings, I thought. As far as I can tell, value and power are about the same thing in the game. Do any of you know for sure?

Alot of the flags in the goals files dont seem to make sense. I got a noticeably more aggressive AI when I increased the threat bonus on siege and attack goals. Altho at that same time I had increased the priorities of attack and siege in strategies, so I'm not sure if the threat bonus alone caused the increase in aggressiveness. But looking at some of the other values, it seems like you get more points for attacking an ally than you do for attacking an enemy unless they are using a double negative. I've ben playing with that chokepointbonus to see if the AI actually can calculate a chokepoint but the results havent been consistent enough to know if anything I'm changing is actually having an effect. I thought I remembered someone claiming that the powerbonus was yet another used flag.

Alpha Wolf,
Thanks for at least putting those extensions back in place, that helps a lot.
Whenever I have some time I'll put these files in a scenario folder and see if they work. If they do, I could put these files on my website or email them to you, 'cause that would basicly be a mod (well, an alpha/beta version at least, I know your still improving things), not much else needs to be changed as far as I can see.

However the diplomatic model in the game is so weak, and the game so tedious, the only think that worked was going to war and conquering.