Why do higher levels curve mathematic ?

[gunter]

Tell me what you want anyway even the younger stones can recognize that combats mathematics have been deeply curved to suit higher levels cheat needings.

Do not tell me I have been suggestioned by a tremendous bad luck sequence,it's seems to me so evident that combat results are dramatically different from Noble vs Prince vs Monarch.

Could you please point me to a mod who prevents once for ever the AI to curve math rules during combats depending on difficulty levels ? I gave up a game cause I started to laugh so much about what the AI was doing that seemed to me a crying child making the whims.

It was definitely more than desperate

 

[ggganz]

Why don't you just not play on higher levels?

 

[gunter]

I am aware of all the " cheats " that the AI deployes in order to win on higher levels but to be honest I wasn't aware that the AI also tweaked the combat math rules.....

It is like playing chess not only with lot of more pieces but also allowing for instance a pawn to move like a tower and sometimes , if really necessary , also like a queen-knight combo......that's crazy !!

That's not fair playing inventing rules on real time

 

[r_rolo1]

^^ A save or a combat log ,please.... if you don't post something more solid, you'll get the " that's a ilusion caused by the human ability of creating patterns where they don't exist" speech.... You should know better than me ( by your join date ) that this kind of statements appear from time to time and most of the times they are simply a bad interpretation of the game mechanics....

 

[ggganz]

They don't tweak the rules of mathematics, they get a combat bonus. They just get another modifier so they have a higher chance.

 

[sylvanllewelyn]

The AI has fixed percentages of increased strength against animals and barbarians. It even took some good players and modders a while to realise, because we don't often visit combat logs of AI vs. animals or AI vs. barbarians. So yes, well observed. I think in BTS it's +40% strength against animals and +25% strength against barbarians, used to be more in vanilla and warlords.

However, if you're just a sore loser with selective memory, then <censored> off.

 

[PieceOfMind]

Gunter are you proverbially pulling our legs?

If not, then using the words of an infamous Aussie pollie, "Please explain."

 

[TommyTankRush]

I think he just means the AI wins more battles than it 'should' mathematically. A variant on the hoary old 'I had 99.7% combat odds' whine.

I recently played a game on deity where I lost 6 battles in a row with 70% odds or better, but that can happen randomly, I know. What I noticed was that I found it very hard to get a unit up to level 6 - which made me think I might indeed be losing troops at a more unfavourable rate.

 

[Psyringe]

Tell me what you want anyway even the younger stones can recognize that combats mathematics have been deeply curved to suit higher levels cheat needings.

Gunter, could you please stop the mudflinging and trolling? The game's combat has been tested extensively by lots of people, and no AI cheat has ever been found. You have no proof whatsoever (actually, you're not even making a clear claim).

It even took some good players and modders a while to realise, because we don't often visit combat logs of AI vs. animals or AI vs. barbarians.

If you're talking about the AI combat bonus against animals and barbarians, then no, these were actually realized the day the game came out. The bonuses are spelled out clearly in the Civ4HandicapInfo.xml, and it really didn't take much guessing to find out what "iAIAnimalBonus" and "iAIBarbarianBonus" meant.

The player does get such bonuses as well. Interestingly, no one ever calls that cheating.

 

[Psyringe]

What I noticed was that I found it very hard to get a unit up to level 6 - which made me think I might indeed be losing troops at a more unfavourable rate.

Well, it *is* hard to get a unit to level 6. Think about a unit that (being careful) is only fighting battles with 90% odds of winning. Such a unit will, on average, survive 10 fights, and will (due to the high odds) never get more than 1 point of experience per fight. So this unit is, on average, expected to die with 10 xp points, which is far away from level 6. You can see that even if you only fight when your odds are very good, you need a lot of luck to bring a unit to level 6 - or you need to be extra careful with your unit. A unit which only fights when odds are 95% in its favor (fo examplee) will survive 20 battles on average.

 

[MrCynical]

Sorry Gunter, but you saying the AI cheats doesn't make it so, especially in the complete absence of evidence. The combat odds have been exhaustively tested, and are accurate (there were small errors in some of the earliest versions, but these have now all been fixed and were nothing to do with difficulty level). If you look through the code, the only modifiers to combat related to difficulty level apply to animals and barbarians, nothing else.

The AI is not a living thing with a grudge against you Gunter. It's just a program following a set of rules and instructions, and those contain nothing to allow it to cheat as you suggest. Numerous tests have shown this to be true, and random unsupported complaints do not change that.

The trouble is that most people do not understand probability - many seem to think that greater than 50% odds translates to "I will always win". Given the number of combats in a game of civ, you should see losses at 90%+ odds quite frequently.

The other issue is good old selective memory. You remember when you lose a 90% plus battle. You rarely remember the occasions when you win a 10% odds battle. I suspect this is where some people get the impression the AI cheats - becuase human players rarely attack when the odds are not in their favour, they rarely see the flukey wins to counterbalance the flukey losses.

 

[L4zXX0r]

Programs cannot cheat. They're programmed to not do that. So if you're AI is cheating, then I'm looking out for Terminators and Neo.

 

[r_rolo1]

Just for the sake of a example, here is the log of 2 turns of a recent SG of mine, involving the capture of a spanish holy city in a hill ( how unusual ). Game is in prince ( we let Toku to govern our empire for 150 turns and we are trying to recover the disaster he created )

Spoiler :

Turn 7/350 (960 AD) [24-Nov-2007 15:27:25]

IBT:
While defending in Spanish territory at Seville, Swordsman defeats (4.44/6): Spanish Catapult (Prob Victory: 97.3%)
Kyoto begins: Catapult (4 turns)
Kyoto begins: War Elephant (4 turns)
Spearman 1 of Kyoto (Spearman) promoted: Medic I
Spearman 1 of Kyoto (Spearman) promoted: Medic II
Catapult 1 of Kyoto (Catapult) promoted: City Raider I
Catapult 1 of Kyoto (Catapult) promoted: City Raider II
While defending in Spanish territory at Seville, Catapult 1 of Kyoto (Catapult) loses to: Spanish Swordsman (4.08/6) (Prob Victory: 69.9%)
While defending in Spanish territory at Seville, Catapult loses to: Spanish Swordsman (4.98/6) (Prob Victory: 64.9%)
While defending in Spanish territory at Seville, Axeman defeats (1.85/5): Spanish Archer (Prob Victory: 61.9%)
While defending in Spanish territory at Seville, Axeman loses to: Spanish Axeman (2.30/5) (Prob Victory: 70.6%)
While defending in Spanish territory at Seville, Axeman loses to: Spanish Axeman (1.30/5) (Prob Victory: 54.7%)
While defending in Spanish territory at Seville, Swordsman defeats (2.04/6): Spanish Axeman (Prob Victory: 56.9%)
While defending in Spanish territory at Seville, Swordsman defeats (1.44/6): Spanish Swordsman (Prob Victory: 30.1%)
While defending in Spanish territory at Seville, Swordsman defeats (2.10/6): Spanish Archer (Prob Victory: 28.8%)
While defending in Spanish territory at Seville, Axeman loses to: Spanish Catapult (1.85/5) (Prob Victory: 27.0%)
While defending in Spanish territory at Seville, Axeman defeats (3.65/5): Spanish Swordsman (Prob Victory: 11.8%)
While defending in Spanish territory at Seville, Swordsman defeats (4.80/6): Spanish Axeman (Prob Victory: 9.6%)
While defending in Spanish territory at Seville, Archer defeats (3.00/3): Spanish Chariot (Prob Victory: 48.0%)
Ivan the Terrible (Great General) born in Kyoto
While defending in Spanish territory at Seville, Swordsman defeats (3.30/6): Spanish Chariot (Prob Victory: 15.3%)
Osaka grows: 7
Satsuma's borders expand
Kagoshima grows: 3
Attitude Change: Isabella(Spain) towards Catherine(Russia), from 'Cautious' to 'Pleased'

Turn 8/350 (980 AD) [24-Nov-2007 15:34:12]

IBT:
While defending in Spanish territory at Seville, Catapult loses to: Spanish Axeman (2.00/5) (Prob Victory: 53.4%)
Kyoto begins: Catapult (4 turns)
Swordsman promoted: City Raider I
While defending in Spanish territory at Seville, Swordsman defeats (1.62/6): Spanish Axeman (Prob Victory: 10.9%)
Swordsman promoted: City Raider I
While defending in Spanish territory at Seville, Swordsman defeats (4.02/6): Spanish Swordsman (Prob Victory: 22.0%)
While defending in Spanish territory at Seville, Swordsman defeats (3.30/6): Spanish Catapult (Prob Victory: 0.8%)

As you can see I've lost some +90% battles and won some 10% ones. In that regard the AI doesn cheat...IMHO, of course

 

[kazapp]

Well, it *is* hard to get a unit to level 6. Think about a unit that (being careful) is only fighting battles with 90% odds of winning. Such a unit will, on average, survive 10 fights, and will (due to the high odds) never get more than 1 point of experience per fight. So this unit is, on average, expected to die with 10 xp points, which is far away from level 6. You can see that even if you only fight when your odds are very good, you need a lot of luck to bring a unit to level 6 - or you need to be extra careful with your unit. A unit which only fights when odds are 95% in its favor (fo examplee) will survive 20 battles on average.

Odds of 90% means that from the perspective of the field commander nine out of ten units will win.

From the perspective of any given unit's commander, however, 90% means that the odds of surviving (at least) ten battles in a row is .9^10=34.9%.

I guess that is what you're trying to say with "Such a unit will, on average, survive 10 fights"; however it's easy to be misled by such a statement - as if the unit's pretty much guaranteed to survive that long, when in reality, it stands a 65.1% chance of dying before its tenth victory.

What 90% means is that you'll on average get nine victories for every loss.

But as dead units don't fight, this fact is much less relevant than the question of getting those nine victories in a row.

The conclusion is that units that you want to survive through an entire game need odds somewhere around 98-99% for you to risk them in combat (each time!) if you wish to keep them with any certainty (after all, the odds of surviving twenty battles at 98% is only 66,8%!)

Best Regards,
kazapp

PS. The above doesn't mean in any way I'm opposing your point. I'm definitely not siding with the original poster. (In fact, I'm not even sure I understand what he's trying to say)

 

[r_rolo1]

^^That is is a completely diferent thing..... I saw a long time ago a post claiming that the RNG generator was the classic " pick a number between 0 and 1 ,square it and remove the non-decimal part. Rinse and repeat." ( can't confirm or refute that ). That particular RNG is well known for creating very long sequences of good/bad luck, and in a game like this, it could mean loosing half of your army in a battle before you scratch the enemies skin.... ( or the other way around )

P.S Someone with Python knowlegle could confirm or deny this?

 

[3 EMS]

Tell me what you want anyway even the younger stones can recognize that combats mathematics have been deeply curved to suit higher levels cheat needings.

Wonder if there is any moss on my butt.

 

[Psyringe]

@kazapp: You're correct, my analysis neglected the fact that dead units don't continue to fight. Seems that my math skills have gone a bit rusty lately. Thanks for putting it right - and for explaining the step I missed so well that there's nothing else I could add.

 

[Todd Hawks]

Hey, it's Gunter again! Funny stuff.

For everyone else: There is no combat cheat for the AI. Period.

 

[Defiant47]

I used to be the biggest anti-cheating candidate... until that one fateful day.

My catapult had 0.1% chance of winning (back when catapults could kill). I did not plan for it to win, but I only needed it for the collateral damage. However, before I knew, it won the battle, and also accumulated a whole bunch of XP. The only explanation could be was that I cheated.

I'm sorry everyone, I guess now I'm going to have to be banned from Game of the Month and stuff... I swear I didn't mean to cheat by having my 0.1% chance to win come true

But what's done is done... and I couldn't keep it a secret anymore.

 

[Tlalynet]

In Guns defence, I (and most players probably) do have a lot more units die as the difficulty level goes up. AI has more troops and beter troops and going to war just plain costs more. It can be frustrating when your used to winning easily with X units and you do the same thing and die miserably.