AI Cheating

He was probably just exaggerating to make a point. After all saying "no war weariness" is a bit faster than saying "severely reduced but still discernible war weariness that is dependent on the difficulty level".

If memory serves WW is reduced by 50% on Noble and 10% for every level above that - this would totally eliminate it at Deity. I don't remember off the top of my head if this is the case or is it just capped at 90%. So not having it would be true after a fashion, but certainly not on Noble ;)

EDIT: Seriously, could you stop repeating your Soren Johnson tagline and actually read what people are writing.
 
its allways cute when someone find a beaten-to-death argument

Accepted wisdom about whats true... a question for you.

How many planets are there in the solar system today, as opposed to a few years ago?

As to the one person, Dave Mark was quoting from Soren Johnson who actually created the Civ 4 AI. If he (Soren Johnson) doesnt know, then who does?

ASG

Could you please shut up about Soren Johnson, and look at actual evidence of war weariness?

I used a few GSpys to investigate one of Ragnars' cities in my previous immortal game:
 
Either the OP didn't look at the pic Windsor posted, or he believes what Dave Mark and Soren Johnson had said over his own eyes. And that, IMO, is a lot of faith.
 
To the OP: you are in error.

The effects of different difficulty levels are widely researched, documented and verified. This issue pops up almost every day as frustrated players lose a game and decide that the AIs had to cheat because they lost to them. And as already noted all credibility your sources had was gone when they claimed something that can be proven false just by looking at the game interface while playing.

frustrated players lose a game and decide that the AIs had to cheat because they lost to them

If you read carefully my original post, you'll see its theme is not winning or losing, but instead that my enjoyment of the game was spoilt by the AI cheating. 125+ turns of the same thing happening is not what I call fun. IF the AI played with the same effects of War weariness, support etc then stupid things like 100+ turns of war would not and could not happen.

I play the game (Civ 4) to enjoy it. To enjoy playing as many races/civs as i can using different strategies. Playing to win isn't my imperative as i dont need to prove myself against a game. Winning is nice to enjoy though as an additional extra as a consequence of playing the game well. Above all its playing and having fun that matters.

125+ turns of attacks seving no purpose is not fun, its a symptom of a problem with the game.

ASG
 
Hold up...

YOU are the one who is questioning the accuracy of my source and not me. Specifically from...

http://intrinsicalgorithm.com/IAonAI/2008/02/gdc-2008-soren-johnsons-lecture-on-civ.html

...Dave Mark says...

But Noble has other cheats e.g.

Animal/Barbarian combat bonuses
No Unit support
Better Unit upgrades
No Inflation
No War Weariness
The AI needs more help in these areas.


If YOU are saying my source Dave Mark is wrong, then take it up with him at the gamedev forum. HE (Dave Mark) is quoting directly from Soren Johnson who you yourself state is/was a lead designer on the Civ 4 AI.
Actually, it isn't really necessary to take this up with Dave Mark because it is trivial. If you are more convinced by words on a webpage than the code that actually runs the game, then I don't think there's any hope for you understanding how the game works.


Now you can't have it both ways.

IF Soren Johnson, who presumably worked on designing, creating and programing the Civ 4 AI says...

I'm not an expert on the make up of game development teams but I can assure you the lead designer is not responsbile for all of those things. Soren did have a large hand in the AI but it doesn't mean that loosely stated things in a lecture given by him are the exact truth. In other words, yes, Soren is lying when he gives that talk if he makes the claims you are saying he made. As I said before, it's ok to exaggerate the truth in a lecture on AI if the exact truth is not terribly important in the lecture. He exaggerated the truth to illustrate his ideas. Do you understand?

Animal/Barbarian combat bonuses
No Unit support
Better Unit upgrades
No Inflation
No War Weariness
The AI needs more help in these areas


...then who am I to believe, him or you? :lol:
Believe him or me, I don't care. But you don't have contact with him do you? If you did have contact with him he'd confirm that you are in error.

If he (Soren Johnson) doesnt know, then who does? ;)

ASG

Um, anyone who has read the code, which is lots of people around here. As I said earlier, the code is accessible to anyone who chooses to read it. How can you say that some written words from over a year ago are more reliable than people who are telling you they have read the code and have verified how it works?

frustrated players lose a game and decide that the AIs had to cheat because they lost to them

If you read carefully my original post, you'll see its theme is not winning or losing, but instead that my enjoyment of the game was spoilt by the AI cheating. 125+ turns of the same thing happening is not what I call fun. IF the AI played with the same effects of War weariness, support etc then stupid things like 100+ turns of war would not and could not happen.

I play the game (Civ 4) to enjoy it. To enjoy playing as many races/civs as i can using different strategies. Playing to win isn't my imperative as i dont need to prove myself against a game. Winning is nice to enjoy though as an additional extra as a consequence of playing the game well. Above all its playing and having fun that matters.

125+ turns of attacks seving no purpose is not fun, its a symptom of a problem with the game.

ASG

May I suggest that your problem here is that since you have learned of the AI handicaps (what you call "cheats") your enjoyment has diminished. I find it amusing the person you are placing so much trust in (Soren) is the one who most likely decided on the extent of most of the cheats. You are complaining about a design decision he made yet using him as your one source of information via a second hand source.

Anyway, if you don't like how the AI cheats then you are welcome to fix the game to your heart's desire. You can mod out all of the AI bonuses if you want. To my knowledge no one has done this yet because it would cripple the AI so much the game would be ruined.

Are you starting to see the light yet? ;)
 
allsirgarnet said:
No Inflation

Inflation Percent as per CIV4HandicapInfo.xml
Code:
	Human	AI
settl	60	100
chief	70	90
warlord	80	80
noble	90	80
prince	95	80
monar	100	80
emp	100	80
imm	100	80
deit	100	80

Do you want to be proved wrong on more counts?
 
Inflation Percent as per CIV4HandicapInfo.xml
Code:
	Human	AI
settl	60	100
chief	70	90
warlord	80	80
noble	90	80
prince	95	80
monar	100	80
emp	100	80
imm	100	80
deit	100	80

Do you want to be proved wrong on more counts?

Infact we are all right...

Civ 4...

Animal/Barbarian combat bonuses
No Unit support
Better Unit upgrades
No Inflation
No War Weariness


Correct as I stated.

Warlords...

AI Bonus
Unit Supply - 10% (Deity) to 50% (Settler)
Unit Upgrades - 5% (Deity) to 45% (Settler)
Inflation - 20% (Deity) to 100% (Settler)
War-Weariness - 20% (Deity) to 100% (Settler)
Barb Combat Bonus - +40% vs barbs and +70% vs Animals


Beyond...

AI Bonus
Unit Supply - 50% Every level
Unit Upgrades - 50% Every level
Inflation - 80% at every difficulty above Chieftain
War-Weariness - 80% at every difficultly above Chieftain
Barb Combat Bonus - +25% vs barbs and +40% vs Animals


Source...

Beyond the Sword AI Programmer - Dated 20th July 2007

ASG
 
No Inflation [/I]

Correct as I stated.

Incorrect. The handicaps regarding inflation are mostly the same in civ 4 vanilla as well (but they get much lower nearer Deity). Specifically, the AIs don't get zero inflation.

It's nice to be able to read the xml data instead of having to rely on ambiguous sources from more than two years ago. :lol:
 
Allsirgarnet your argument from authority is a logical fallacy. Just because the lead designer says something doesn't make it true. By being able to directly observe code and subjectively witness outcomes that disprove your statement we can easily prove that your quote is, in fact, at least misleading if not outright false. Your reliance on a mystical trump card based on a logical fallacy makes you look ignorant and foolish.

It is plan to see in the code of the game we are discussing that you are wrong.
 
Infact we are all right...

Civ 4...

Animal/Barbarian combat bonuses
No Unit support
Better Unit upgrades
No Inflation
No War Weariness


Correct as I stated.

Warlords...

AI Bonus
Unit Supply - 10% (Deity) to 50% (Settler)
Unit Upgrades - 5% (Deity) to 45% (Settler)
Inflation - 20% (Deity) to 100% (Settler)
War-Weariness - 20% (Deity) to 100% (Settler)
Barb Combat Bonus - +40% vs barbs and +70% vs Animals


Beyond...

AI Bonus
Unit Supply - 50% Every level
Unit Upgrades - 50% Every level
Inflation - 80% at every difficulty above Chieftain
War-Weariness - 80% at every difficultly above Chieftain
Barb Combat Bonus - +25% vs barbs and +40% vs Animals


Source...

Beyond the Sword AI Programmer - Dated 20th July 2007

ASG

I did difficulty level modding back in vanilla - I made a version of Noble where everything was equal between AI and the player. I can confirm that there was inflation for the AI in vanilla in Noble, at least (I think it was in all difficulties, though).
 
Stacking the deck for the AI and against the human player at higher levels has been the common practice for games since Doom. To have the AI actually play a more sophisticated game as you go up in difficulty would require immense amounts of coding as well as substantial brute force computing power. In other words; if you want something more than AI "cheats" you'd have to be willing to shell out a few hundred bucks for the game and also buy a computer that would make NASA green with envy.
 
The only strategy game I can think of that had the computer actually play smarter at higher difficulty levels was Galactic Civilizations 2. They coded a really smart AI that took into account all the different ways to win and used some interesting strategies, and it was very difficult to beat. On lower difficulty levels they coded the AI to deliberately make mistakes, overlook things, and not use certain effective strategies.
 
Stacking the deck for the AI and against the human player at higher levels has been the common practice for games since Doom. To have the AI actually play a more sophisticated game as you go up in difficulty would require immense amounts of coding as well as substantial brute force computing power. In other words; if you want something more than AI "cheats" you'd have to be willing to shell out a few hundred bucks for the game and also buy a computer that would make NASA green with envy.
No, not really. Your storage would have to be high but terabyte harddrives are cheap nowadays. It would be impossible to create An AI that is sophisticated enough to play as a real player. We just do not have the technology to do so because AI's think in absolutes: 1/0, on/off, yes/no. In other words binary. We do not, we think in trinary: 2/1/0 or yes/maybe/no. Because of the limitations of binary absolutes, we must program around them to give an approximation of choice. If we were to design programing in trinary, then we would create an AI that would go insane or just stop working because it would be confused or unable to make a choice. It would also introduce mistakes made by the AI on the level that humans make every minute and would make them just as untrustworthy as humans. Instead of removing the human element you would just match it. That said, just play the game and if you get into a stalemate, revert save and try a different strategy. The AI will never be perfectly human like, so some elements have to be added to even the score. We can multi task and they cannot. Programs can do a close approximation of thinking backward but cannot think forward. So every decision it makes is based on your current choice. It cannot guess what your strategy is so it is quite limited in its ability to project outcomes, without weighing the choices it has in it's limited table. We do not have these restrictions.
 
This is the most fascinating argument I've seen in a while. I don't see the point of it, but it's fascinating, nevertheless.

Stacking the deck for the AI and against the human player at higher levels has been the common practice for games since Doom. To have the AI actually play a more sophisticated game as you go up in difficulty would require immense amounts of coding as well as substantial brute force computing power. In other words; if you want something more than AI "cheats" you'd have to be willing to shell out a few hundred bucks for the game and also buy a computer that would make NASA green with envy.

I've always wondered what it would be like to play Civ against a super computer (i.e., those computers they pit against chess champions). It'd probably be super frustrating.
 
]I play the game (Civ 4) to enjoy it. To enjoy playing as many races/civs as i can using different strategies. Playing to win isn't my imperative as i dont need to prove myself against a game. Winning is nice to enjoy though as an additional extra as a consequence of playing the game well. Above all its playing and having fun that matters.

If you just want to have fun rather than prove yourself, why not drop back down to warlord? That way the bonuses you recieve will offset the AI edge. I mean, if it really upsets you this much...

Or you could edit the XML, it's not like you need a degree to do it. I edited out these advantages 3 or 4 years ago and it really hurt the AI way too much to be fun. While you're at it, don't forget that the AI cheats by having 10 hammers in the queue when they found their first city. (another known non-issue around here)

Of course, the idea is to use the unfair advantages YOU are supposed to enjoy (i.e. a brain) to offset these advantages the AI get.
 
Allsirgarnet your argument from authority is a logical fallacy. Just because the lead designer says something doesn't make it true. By being able to directly observe code and subjectively witness outcomes that disprove your statement we can easily prove that your quote is, in fact, at least misleading if not outright false. Your reliance on a mystical trump card based on a logical fallacy makes you look ignorant and foolish.

It is plan to see in the code of the game we are discussing that you are wrong.

All my info is based on statements made by programmers who actually worked on the game(s). As to a mystical trump card once again my point is missed, in that in my experience of playing the game it is often made worse by the cheating.

Unlike many my world doesnt either revolve totally around civ or more importantly, revolve around BEATING civ at high levels... as shown by some members here posting info on their profiles regarding civ.

Probably whats narked many members here is my apparent belittling of a sacred cow, which in history always gets the mob onto the streets for a good burning or hanging. ;)

In my long history of 4 decades of gaming of all types, ive found theres far more fun in just playing the game over winning, even though in the latter i've done pretty well. As a board game designer i've found that fair play and fair rules are important to nearly all gamers, and yet in many cases this doesnt apply. What matters to many is winning, or rather beating the game (such as civ) to prove themselves in some way. The problem with the latter is that its just a game, meant i presume to be played for fun.

From the many posts ive read in many threads here and from quite a few profiles, things have gone way beyond fun into the areas of either obsession or even fetish.

That friends is not healthy. ;)

As to many of the responses here they were way over the top in many cases, especially as my post was meant be both ask questions and hopefully be informative. Sadly many of the responses seem to confirm many of the accepted predjudices common about forums on the net.

Civ in all its versions is just a game. No more, no less. Its not some sort of god to be worshipped at a forum altar.

As to the cheating itself, its been in my experience far more common than many believe in very many games. I well remember Microproses 'Master of Magic' which not only had Civ like cheats, but actually allowed the AI to break the game rules in virtually every area.

Perhaps whats needed is an upfront cheat rating published (like the minimum specs) with the game. Then a player can adjust his or her perception or view of the game before they play it.

Finally can i mention the board game Illuminati as deserving of an award. It unlike most every other game does allow and encourage cheating as stated clearly at the end of its rules. :goodjob:

Regards

ASG

Moderator Action: Trolling - warned
Please read the forum rules: http://forums.civfanatics.com/showthread.php?t=422889
 
If you just want to have fun rather than prove yourself, why not drop back down to warlord? That way the bonuses you recieve will offset the AI edge. I mean, if it really upsets you this much...
Every so often I get tired of proving myself. The amount of thought and knowledge necessary for me to struggle at higher levels and the micro required to get the best results from them becomes momentarily tedious. Rather than dropping down a level or two I just load the Full of Resources map mod and then alter the settings so that all the civs get as many resources as will fit. Then I set it for Custom Continents so that each civ has it's own. One game on a fantasy map like that and I'm ready to go back to losing on regular maps.
 
On the subject of AI...

I've sometimes considered that one of the problems with AI in games is the nature of AI itself, in regard of the programmers who create it. Now as I see it in simple terms AI is used to allow or create (in Civ for example) enemy computer players, and it dictates or drives how those players... play the game against you the human player. This then places the programmers (in effect) as the enemy players, for they actually determine the strategy the computer uses via the AI to play against you.

My problem is this...

IF you accept the above as fair and reasonably accurate, then surely the overall quality of what the AI actually does depends in the main, on how good a strategist or player the programmer is. By this I mean in creating the stategies the AI uses to play, the programmer surely must put their own strategic or gaming skill in the mix.

This my friends is the point where i have trouble with cheating in that cheating possibly might be used to make up for the poor gaming or strategic skill of the programmers. I might go further and say that possibly in addition to making up for lack of skill, the programmers might use cheating simply out of pure laziness in that it saves programming time and effort... or rather it saves them work.

Think on it...

How long would it take to create a half decent non cheating AI, as compared to just cheating be beefing up the computer players in all areas. That has always rankled me a little as a game designer whos creations (played by only humans) have to be totally fair.

Perhaps the theme of cheating, laziness or otherwise stems back to its precurser boardgaming, where certain companies (SPI being a prime example) used to pump out endless half finished games that only became playable when the players finished the rules themselves.

A last titbit...

Between boardgames and PC games these days there used to be (and still is) PBM or play by mail games, where the players sent in turns to be moderated and processed by a computer at the owning/running company. One such game I played literally 20 years ago now was QUEST, a fantasy strategy/semi RPG war game. With a group of friends I played the game, which cost a weekly fee, and we explored every part of the game.

Certain aspects such as inner temples always eluded us despite our best efforts. Then one day we visited a games convention in Bethnal Green in London and spoke to one of the game designers at a Quest trade stand. My friend Chris Wilson, who was a long standing and well respected play-tester of games asked about the 'inner temples' and other aspects that eluded us. Here was the gist of their reply...

We originally designed the game to included functioning inner temples and the like, but never got around to actually creating the game mechanisms to include them. So although they are in the game, they serve no function. :eek:

We all left the game within a few weeks in disgust. ;) Perhaps thats where my cynicism and distrust in part stems from. ;)

Anyway, I hope my musings on things AI have been of some interest.

Regards

ASG
 
All my info is based on statements made by programmers who actually worked on the game(s). As to a mystical trump card once again my point is missed, in that in my experience of playing the game it is often made worse by the cheating.

Unlike many my world doesnt either revolve totally around civ or more importantly, revolve around BEATING civ at high levels... as shown by some members here posting info on their profiles regarding civ.

Probably whats narked many members here is my apparent belittling of a sacred cow, which in history always gets the mob onto the streets for a good burning or hanging. ;)

Let's see... You start a thread which contains blatant misinformation, and are objectively proven false in every relevant way. You however fail to acknowledge this and cling to a single irrelevant and outdated comment (backed by some ambiguous outdated references) by a game developer. This is what "narked" many members - it has nothing to do with the game, rather the feeling of having an argument with a brick wall. I didn't really even understand the sacred cow analogy, usually the more people play the less glorious is their view about the game as they bump into all the small things that are poorly implemented. If people thought the game was somehow holy wouldn't they consider Soren Johnson a god and heed his words faithfully? ;)

Now you proceed to take the high and mighty attitude of "at least my life doesn't revolve around the game". While that's a common stance taken on Internet forums by people that don't want to admit they are wrong, it's a pretty immature one and I'm surprised to see someone with 4 decades of gaming history resort to that.

In my long history of 4 decades of gaming of all types, ive found theres far more fun in just playing the game over winning, even though in the latter i've done pretty well. As a board game designer i've found that fair play and fair rules are important to nearly all gamers, and yet in many cases this doesnt apply. What matters to many is winning, or rather beating the game (such as civ) to prove themselves in some way. The problem with the latter is that its just a game, meant i presume to be played for fun.

From the many posts ive read in many threads here and from quite a few profiles, things have gone way beyond fun into the areas of either obsession or even fetish.

That friends is not healthy. ;)

For many people "fun" means just playing the games, paying no heed to winning or losing. For many others "fun" means getting challenged by the game and still being able to beat it. You seem to indicate the latter is somehow automatically unhealthy, why is that? And how on earth can you tell by someone's profile if they're having fun or not? Or being obsessive?

As to the cheating itself, its been in my experience far more common than many believe in very many games. I well remember Microproses 'Master of Magic' which not only had Civ like cheats, but actually allowed the AI to break the game rules in virtually every area.

Perhaps whats needed is an upfront cheat rating published (like the minimum specs) with the game. Then a player can adjust his or her perception or view of the game before they play it.

Yes, cheating is (at least historically) common for programming AIs, because of the sheer complexity in programming a competent AI in a complex game. However, Civ4 (BtS, I've understood that the vanilla AI is much much worse) AI is among the top ones that I've seen in any game, and doesn't actually "cheat" in any real sense (there are some minor things of which many are mentioned in this thread) - there are handicaps which are quite similar to the handicaps given to the human player on low difficulty levels. Is the human player cheating if he's playing on Settler?
 
This is the most fascinating argument I've seen in a while. I don't see the point of it, but it's fascinating, nevertheless.



I've always wondered what it would be like to play Civ against a super computer (i.e., those computers they pit against chess champions). It'd probably be super frustrating.
It is actually the program and not the computer you are playing. We think of Chess as being separate because there is a physical board (dedicated chess chips) involved but the reality is, the only thing missing is the GUI (graphical interface). I probably did not explain that very well.

With Deep Blue and Kasperov, the decision making was more in depth only because the programmers studied the famous moves and counter moves of the masters and in fact got two grandmasters to help in the programing. The computer really did no learning of its own accord and if fact, between games certain upgrades were made to programing to avoid pitfalls where Deep Blue was demonstrating weaknesses in earlier matches. Modern programs and Hardware is much more powerful now than it was hen and a standard laptop has more computing power than Deep Blue. Since deep Blue was designed with dedicated chips for Chess, it would be a poor competitor for CIV.
 
Top Bottom