# The AI is booooring

Discussion in 'CivRev - General Discussions' started by Ontolog, Oct 6, 2008.

1. ### SmirkKing

Joined:
Dec 2, 2001
Messages:
839

You are confusing dinner with the plate. The fact that computers and programming is logic has nothing to do with the 100% mathematical properties of games like chess. As another poster mentioned chess is 100% player controlled, this means that I can predict with certainty what you will do in any given turn and determine with 100% accuracy what the outcome of those choices will create. You will not attack a pawn with your queen and fail.

Even in chess you have an incredibly amount of complexity, and this increases as the game progresses (but decreases as it winds down). But Civ and other games of that nature are on a different level. Thats not to say it can't be improved, I'm sure it can be, but AI is not "easy". I'd rather be able to play friends online, or even hotseat play.

As to your first point above, this may just be semantics. What constitutes logic when your decision is based on a probability? You can't read your opponents mind so you are just guessing what he may move based on the value of different moves and perhaps the personality of the player? At any rate, as that applies to computers and computer AI means its just math. I'm not up with the latest in chess AI, but generally in the world of AI the computer is doing a lot of calculating, its not thinking, nor is it using logic.

Your second point I was referring to above with the dinner/plate analogy. But while this line of code "x=randomNumber(0,100)" may be 100% logic programatically, the outcome and the value in x after it completes is random.

2. ### HertshDarwinist

Joined:
Jul 18, 2008
Messages:
311
Location:
Stockholm, Sweden
You are right there. The random part of an attack in Civ is different from Chess and thus makes it more complex. However, there is probability of an outcome and not just full random outcomes. Still you can optimize the AI decisions based on the probability of all potential actions a player can conduct. I admit it is much more to optimize than on an 8x8 board, but still possible.

However, I think the two of us have made our point and we are basically trying to split hair. Or?

3. ### JayAhre 13Chieftain

Joined:
Jul 9, 2008
Messages:
5
Location:
Fort Worth, TX
you're missing the point. Civ does not equal chess. It cannot. In chess, there is no random outcome. In Civ, there is. Civ /= Chess

4. ### OntologChieftain

Joined:
Oct 2, 2008
Messages:
13
He is not missing the point. He clearly stated "However, there is probability of an outcome and not just full random outcomes. Still you can optimize the AI decisions based on the probability of all potential actions a player can conduct."

Computers can do statistics much better than we can. There just needs to be a well designed engine it should be superior to a human player.

The main difference between Civ and Chess with regards to AI is that incredible amounts of money and research have been put into figuring out how to make a decent chess engine over the years. Civ has not had this pleasure so its AI is pretty naive compared to your typical chess AI.

5. ### HertshDarwinist

Joined:
Jul 18, 2008
Messages:
311
Location:
Stockholm, Sweden
I could not have said it better myself.

6. ### SmirkKing

Joined:
Dec 2, 2001
Messages:
839
You're still wrong however. There is a significant difference between Civ and Chess. Unless you've thought about this in terms of programming AI systems you can't really grasp the difference that enters with something as simple as a bit of randomness. I'm not talking about statistics, sure Chess AI need to understand probability, they weigh the outcome of every possible move, but thats just not possible in Civ.
Its even very difficult in a simple game like Backgammon, compared to Tic-Tac-Toe or Checkers which are simple but trivial to code an AI. Even in Backgammon though, the play is very determinant, this is not so with Civ.

7. ### OntologChieftain

Joined:
Oct 2, 2008
Messages:
13
For outcomes that are truly random, computers and humans are on even grounds guessing which is the best course of action, they can both 'flip a coin.' However where the outcome of situations (such as a battle between two units/armies) can be reasonably judged by a human, the AI can also make a reasonable judgment, I would even say a better judgment than the human in the case of a Civ game. This is because the same programmers who have written the rules for determining the results of clashes can contribute to the code for the AI. The AI will know the exact set of rules that will be followed when determining the outcome of a battle and will be able to make a decision based on this calculated probability of winning. The human however can only glean what he can from the documentation and educated guessing and even then can't calculate everything on the fly the same way a computer can.

8. ### MichelassoChieftain

Joined:
Oct 2, 2008
Messages:
12
Location:
Treviso, Italy
I don't know why, but I am pretty sure that the randomness and the fog of war, which introduce a luck factor, makes the development of a true AI for CivRev much easier than the one for Chess. At the end it should just be a matter of a good set of decision rules, something that can't work fine with Chess.

9. ### Padmathe InbondAdministratorSupporter

Joined:
Dec 10, 2001
Messages:
14,408
Location:
Omaha, Nebraska USA
One of the developers at Firaxis said, a few years ago, that writing a Civ AI that could beat the pants off the human player was relatively easy. It was writing one that would 'challenge' the human player that was tough....

10. ### Lone WolfDeity

Joined:
Dec 4, 2006
Messages:
9,901
I don't think that guy meant that it is possible to write an AI that with no bonuses of any kind over the player and with no anti-human bias will always outperform the player.

11. ### alcalsurfin'

Joined:
Jul 1, 2008
Messages:
809
Location:
Miami beach
Isn`t logic linked with mathematics?

12. ### EndurePrince

Joined:
Mar 23, 2008
Messages:
307
Location:
Canada
Then why, even going back to a game with a pretty good AI (civ 4) does the difficulty level just change the amount of tech/production/etc bonuses the player or the AI have relatively to each other?

Civ Rev's AI is horrible compared to civ 4, there's really no contest between the two.

I have no doubt whatsoever the developers could give the AI tech/production/etc bonuses to the point where the AI would simply overwhelm the human player, regardless of what they did. However, making a smarter AI I remain very doubtful of since a) Civ Rev's AI seems to be such a regression from Civ 4's AI, and b) since so far, they only make the game harder/easier by changing the bonuses and not by improving the AI's performance itself.

13. ### bob_xWarlord

Joined:
Sep 9, 2007
Messages:
110
Location:
Manhattan, NY, USA
They've been working on chess engines for over 20 years. And chess is a game with much simpler rules: only 64 squares and 32 pieces.

Every version of Civ is a completely different game, even though they my look similar, so the AI has to be rewritten from scratch. The goal is acceptable AI, and not the ultimate AI challenge.

That said, i'll stick with Civ4 BTS and not buy this DS version.

14. ### LordGekPrince

Joined:
Aug 22, 2002
Messages:
520
It isn't Civ in any traditional sense and the AI is lacking, but it's still a fun little Civ Lite, if only the DS version so you can play something in the flavor of Civ all portable.

I started with the DS version but HATING the idea that there was additional content released for the game I couldn't get without having one of the big consoles I just bought a XB360. So I guess I have to talk it up just to justify my \$400+ spent for this game (DS version, XB360, XB360 CivRev game, Guide, and additional map/scenario packs).

15. ### dirtwormWarlord

Joined:
Jul 12, 2004
Messages:
146
Location:
Pennsylvania
well after this thread i am not going to be buying this game.
i was given the idea that the AI has pretty good, but obviously not!
thanks for saving me 60\$!

16. ### LordGekPrince

Joined:
Aug 22, 2002
Messages:
520
I think it improved in v1.3, the AI now willing to fight other AI folk more when appropriate and able to pull off better multi-pronged attacks...but I gather it still cheats like a son-of-a-. .. .. .. .. . in the Emperor and Deity difficulties.

17. ### ncwarriorRuler of my room

Joined:
Nov 14, 2008
Messages:
56
Location:
Antwerpen, Belguim
i completely agree with you there should be some big update that could fix it and then the game could really become fun because it's fun to play but the AI actions make the game less fun especially when your nations are doing good together and the next turn you're in full war with them

18. ### LordGekPrince

Joined:
Aug 22, 2002
Messages:
520
I see your point but I quickly realized as I let my settler wander by the Mongols that I thought I was on good terms with, the AI is backstabbing cheese. At the same time I also get that this is somewhat intentional since diplomacy/relations are fundamentally different in this game than in PC Civ. In CivRev there is no real peace or alliances, you are ALWAYS at war or the brink of war and can trust no one. It even sort of flows from the attitude of the over the top leaderheads, you play one of 5 power hungry paranoid twits setting about to DOMINATE THE WORLD and even if at peace with another Civ it is only a ploy to hold them off until you can finish them off.

So while the AI isn't great and cheats like a bastage, some of the AI's seemingly psychotic behavior makes sense if you see it from the perspective that they are ALWAYS scheming against you no matter what they may say.