AI cheats with battle??

[KatanKing]

first : Civ is the best game ever, have played it thousends of hours since civ1.

second :the only real big completely strange (and stupid) aspect is the unfairness in battles. This is the only big thing that HAS to be improved in civ4.

I've had lotst of combat situations where there was a 100 % defender win (according to the combat calculator), and still lost 5 times in a row.

The frustration is big when attacking a regular rifleman with a veteran pantzer (80% chance of winning, I've calculated 80% chance of losing.. ) and the next turn I loose 4 veteran mech.inf. to stupid AI-Cavalry, and it wasn't even close.

I've had troubles attacking a spearman with tanks, I'm not trying it with Cavalry anymore..

Everybody who plays this game must feel the same, I really don't understand how they could make this aspect so unfair and unrealistic.

question1: am I the only one?
question2: can someone tell the civ-team or are they blind??

 

[citizen001]

it's all in the random dice roll calculation thing, as long as the defender has even 1 defense, and 1 hp, it has a chance of winning. that may explain the cruel losses you're suffering. firaxis could make it so that a unit of one age (ancient age) cannot defeat a unit of another age (modern age).

 

[bingen]

question1: Yes and no. Lots of people feel the way you do, the spearman vs. tank issue is so classic it has gotten its own icon: . However, the PRNG(pseudo random number generator) in Civ has been tested rigorously and shown to be unbiased. On the other hand, the human mind is not biased.

Also, your combat calculator might be off, there is never a 100% win percentage.

 

[Tatran]

Use more artillery units.
Units with only 1 hitpoint are easier
to kill than units with full health.
Even a healthy horseman has a good
chance to win against a rifleman or
guerilla with only 1 hitpoint.

 

[rubixd]

There should be an attribute called "armored"
and it prevents any unit built before the musketman from even damaging the armored unit.

 

[Gogf]

I have better luck attacking spearmen with archers, swordsmen, and medieval infantry then knight and cavalry .

 

[Tomoyo]

That would be because you expect the Knights and Cavalry to win, while when your Archer kills a spearmen flawlessly, it is a very nice surprise.

 

[Padma]

:sigh: Not this again.

The is no cheating in the combat resolution algorithms. The pseudo-random number generator used conforms very well to all measures for randomness. But true random numbers are much more "streaky" than the average human thinks. Plus, the human mind is geared to recognize patterns, so we see patterns in the RNG even when there aren't any. Finally, we remember the *bad* things that happen to us, and they stick out in our memories, but we forget all the times when the opposite occurred, and we *benefited* from it.

To sum up, the AI doesn't cheat in combat. It is strictly a human perception.

 

[punkbass2000]

:sigh: Not this again.

The is no cheating in the combat resolution algorithms. The pseudo-random number generator used conforms very well to all measures for randomness. But true random numbers are much more "streaky" than the average human thinks. Plus, the human mind is geared to recognize patterns, so we see patterns in the RNG even when there aren't any. Finally, we remember the *bad* things that happen to us, and they stick out in our memories, but we forget all the times when the opposite occurred, and we *benefited* from it.

To sum up, the AI doesn't cheat in combat. It is strictly a human perception.

Thank you, my thoughts exactly. I was a little apprehensive about posting 'not this again' to a newbie, though.

 

[Padma]

"Not this again" was partly in response to his 2 questions at the end:

question1: am I the only one?
question2: can someone tell the civ-team or are they blind??

Just a way to be sure he understood that this *has* come up before (again, and again, and again, ...).

I hope KatanKing doesn't take it in the wrong way. If he does, I apologize. It is just that some of us "old farts" here have seen this so many times before, but it has been a while since there was a thread on it.

 

[Grogs]

I once read a very eloquent article on the difficulty of generating 'random' numbers on a computer in a forum for another game. As Padma eluded to, the seeds used for generating random numbers tends to make them a bit streaky. Since there's no way (that I know of) to exploit this streakiness, it all balances out. Two nights ago, I went 7 of 7 with cav vs. musketmen inside of a city. It was streaky enough that I noticed it rather than just saying 'yeah, I'm awesome.'

That being said, even things that should be truly random, like dice, can be very streaky. Being an old paper and dice RPG player, I've seen people roll 4 natural 20's in a row (on a 20 sided die) and then roll 3 in a row on the next round. These are some of the most superstitious people you will ever meet and just about all of them have 'lucky' dice.

 

[shadow-walker_]

:sigh: Not this again.

The is no cheating in the combat resolution algorithms. The pseudo-random number generator used conforms very well to all measures for randomness. But true random numbers are much more "streaky" than the average human thinks. Plus, the human mind is geared to recognize patterns, so we see patterns in the RNG even when there aren't any. Finally, we remember the *bad* things that happen to us, and they stick out in our memories, but we forget all the times when the opposite occurred, and we *benefited* from it.

To sum up, the AI doesn't cheat in combat. It is strictly a human perception.

that is a lie i know for a fact that the zulu archer uses aimbot

 

[Yuri2356]

What's so special about Zulu archers? Babylon has the archer UU.
Zulu Impi used stat-hacks to defeat British.

 

[shadow-walker_]

well in my games the zulu has a regular archer down to one life that attack me and kills my elite archer that is full life. i say BS the computer does it 3 more times and then im screwed and i have to make a new game because they go war alliance vs me with india and IMHO war elephants < france.

 

[King PJ]

Perhaps maybe someone could make a mod that ups the attack and defense numbers between units from different ages? This would certainly help tanks to defeat spearmen yet still be on par with other units from the industrial age.

 

[Tomoyo]

Or you could lower the randomness of the RNG by tripling the HP of every unit...

 

[King PJ]

But then battles would last forever! I run a Pentium 4 2.6GHZ with 256 MB RAM and when I get into the Industrial Age it takes one to two minutes per turn during a war. Tripling the HP of every unit would certainly make it slower.

 

[IbnSina]

I have always thought that the problem is that the RNG uses (pseudo) uniformly distributed random numbers. I think we would not have to revisit this issue repeatedly if Firaxis would run those through a filter to make them normally distributed. If they felt that took too much computation time, there are other things they could do to make the central tendencies of the distribution more pronounced. The results we would get from that would conform much more accurately to our expectations.

 

[warpstorm]

If they were normally distributed then stronger units would almost always destroy weaker units. This is not Firaxis' goal. They wanted lower tech/resource units to have a slight chance of winning any battle.

 

[IbnSina]

They would still have a slight chance - it would just be more slight than it is now