AI cheating in combat?

[Gen.Dragolen]

AI Winning Streaks

Well it's nice to see I'm not the only one who's had problems dealing with the AI's luck in battle.

After getting 9 swordsmen killed in 2 turns by two impi and a warrior (3 on my attack, then 3 on their turn, 3 more my next turn) and watching them go from regulars to elites, I decided it was time to reinstall the game.

In the first game after the reinstall, it appears that things are back to a more normal distribution of luck: took 5 swordsmen and 2 archers and reduced two neighbouring civ's to 1 city each. Granted I made sure neither got even a sniff of iron.

It seems that the combat engine is definitely setup to always give a small chance for a defending unit to do damage no matter how bad the mis-match. I can remember to vividly watching tanks get wasted by cavalry while the AI was fighting a war and I was ref'ing it.

If the basic chance of doing damage is as the manual sets out

attacking unit's attack strength - A
defending unit's defending strength - D

The chance for one unit to damage the other for the attacker

X= A/(A+D) so for the cavalry X= 6/(6+8)=0.42857 or 43%

and for the defender

Y=D/(A+D) for the tank Y=8/(6+8)=0.57143 or 57%

So that makes the tank's chance of surviving combat pretty bad each turn it's in combat. Especially in the face of not being able to retreat when down to 1 hp.

What they should have done was give the tanks and other modern units more hitpoints if they were going to have a over simplified combat system like this.

If reality worked like this the Polish Cavalry would have massacered the German Army in WW2.

[Lush]

I can't really say if the AI "cheats" or not, but I definitely think that the combat system in Civ3 stinks. Even if the combat system is completely fair, the amount of randomness makes the game extremely frustrating.

It's especially frustrating when you like to play aggressive or have no choice because every five turns one of the AI civs declares war on you for not turning over some technology and half your treasury for free.

[notyoueither]

Originally posted by teknokrat

The problem I see is not so much spearmen taking out armour ( I can live with that) but what appear to be "runs" when a single unit always "wins". Personally I think the combat engine has some kind of bug in the way it calculates probabilites.

later

I'd bet you would have been pretty p*ssed as a German when the British escaped at Dunkirk.

Then again you would have been p*ssed when the Soviets held during the early stages of Stalingrad.

And again when the Airbourne (101st?) held Bastogne.

It sux to be a German.

Is it possible that the random generator has a *miracle* switch/string. One that says that any given unit may hold out against all comers on a given turn? I don't mind it, but I will admit I have seen through the eyes of those German generals and not felt very calm about it. Then again, I have seen one of my Spearmen win against 16 Barbarian Horse on a single turn, so I won't b*tch too much about it.

Salve

[jan3]

yeah seems that the luck factor is just an other word for cheating in favor of the ai
since the player is never lucky or very rarely
and if the ai is in bad situation he gets often the help of a major barbarian uprising
i know just playing with the knowlegde that you need 5 mounted warriors to kill one jag.warrior hurts a little or are there logic explanations to this?
cant really see any
and the posts i saw until now.....nothing really convincing

[rid102]

Then again, I have seen one of my Spearmen win against 16 Barbarian Horse on a single turn, so I won't b*tch too much about it.

That's because the player gets a big bonus against barbarians on all but the hardest levels.

The "problem" with combat is the number of hitpoints is basically too low.

Lower numbers of hitpoints => combat is more prone to randomness

Higher numbers of hitpoints => combat averages out over HP depending on A/D.

So having low hitpoints means the combat result is essentially pretty much entirely random; higher numbers of hitpoints mean that combat becomes (nearly) entirely deterministic.

[Arrian]

The AI does not cheat in combat. The wacky combat results that occur are a result of the combat system in Civ III, which has no firepower, and if you don't like the way it works, I understand. It can be frustrating. However, it works for you to, so it is not the AI cheating. Sometimes YOU get the "miracle" result.

For example, last night I was playing around with the Iroquois (Monarch level) and was fighting the Americans. It was late and I was just messin', so I was being kinda careless. I left a veteran Mounted Warrior out in the open. A veteran American archer attacked it - 2 attack versus 1.1 defense, even hp. My MW didn't lose a hitpoint and was promoted to elite. Did I cheat? No, it was just a wierd result of the combat system.

-Arrian

p.s. Teknokrat: what in the WORLD were you doing using riflemen as attack units???

[vmxa1]

Yes it is the lack of FP and the low HP that create the strange combat results. You can have Calv attack say an archer and lose. If you reload and have some other unit fight and then come back to the Calv, it may win and not lose a HP. This is a bit weak to me. I do not mind a little fudge factor for luck, but to go from a stronger unit losing to not taking any damage is too wide of a swing. I am not sure that a doubling of the HP's will make it go away. FP is needed to get it more in line with expectations. I am not avocating that a stronger unit never lose, just make very rare and maybe never for real mismatches, say Knight vs conscripted warrior or BB vs galley. Spearmen vs armour. Those I would not want to see win more than once in my life.

[notyoueither]

Originally posted by rid102


That's because the player gets a big bonus against barbarians on all but the hardest levels.

It was on Regent or Emperor difficulty. Furthermore, I have had my own units hold out miraculouslly against AI units of good quality on other occassions also.

The point remains that the miracles happen for the player as well as the AI.

[jgflg]

Within a small sample of events, you're always going to see streaks. For example, if you flip a coin an infinite amount of times, it'll land 50/50 heads and tails, but if you take any group of 100 flips from that infinite amount, you will always see streaks of all heads or all tails.

[bigphesta]

ok... I think this is what the programmers were going for in this system...

Most wars 'last' a while. Armies dont' just walk to a city and take it. Battles were fought, armies had stratagies, there were winners and loosers, but it wasn't just a bring army A to city B fight for X days and take it over. There was some 'time' to actually control and maintain control of a city. If you're attacking a city, and you're seing you loose guy after guy ( not to pick on anyone, but some people have stated that they lost 10 guys in one turn trying to take a city ) then WAIT and attack next turn. Obviously the 'code" is preventing you from just walking over and dominating the city in one turn. So instead of attacking, move your guy to the next adjacent square and look it as 'getting a flank" on the city. If that guy is moved, it seems that the 'ranking' is reordered and you can win.

Basically, If you loose 2 guys strait, quit attacking. Regroup and come back next turn. You'll have better chances and it will feel more realistic.

Of course, I realize that the defending city "might" be able to create another defender or move in some reinforcements from other areas... But you were going to have to kill them anyways, this just makes it easier later... and you'll have one turn to get more attackers ready as well.

Just my $.02

[jan3]

yep you are right i got i few very lucky victory's to....but still it's weird
maybe because we are to much counting on the results of civ2 where it happened less often.
but still it is a great game (how often i dident say i would uninstall it and 30 mins after started a new game)

[vmxa1]

Bigphesta, we understand that you can attack on a different order or turn and get different results, what I want to see if a lot less of that. I am fine with complete randomness for more or less equal units. It is fine if the amount of damage varies when stronger units fight weaker units. It is not fine when hopelessly out powered units win under any senario. This is all done to prevent use of reload and change results, that is fine, but the random seed is too great of an impact. Fire Power ratings would make that less of a problem and tanks would no longer lose to warriors. All the talk about cities and such is crap. Gallies should never beat BB, never never never. It is debatable if they should even do any damage.

[barefootbadass]

Originally posted by vmxa1
It is not fine when hopelessly out powered units win under any senario.

This has never, ever, EVER happened to me(sure I've lost an individual unit now and then but I've never had an invasion fail against inferior defenders(because of those defenders)). Are you doing something like attacking a city with three defenders with only 2 tanks? The ai can get new defenders to replace those it lost if you don't get it in one turn. The absolute minimum you need is equal to the number of defenders, and you should have probably twice that anyway(plus artillery). Nothing wrong with that, the damage ratio will turn out right in the long run. Absolutely nothing wrong with how it works now. Your mobile units even have a chance to survive if they can't kill the defender.

[vmxa1]

By any scenario, I mean one on one. Of course if you sent two tanks to a city with 6 defenders, you are not going to take the city on that turn. I mean that my tank should not lose to a warrior, even if it is in a city or even a metro. Actually I do not get upset if I attack a city and lose any battle, it is on open fields that I get mad. I do not accept a symbolic representation that is being passed off as a way it could happen. As if the warrior really represents some thing other than a warrior, such as a green beret or navy seal or some such crap and has modern tools. It is a warrior with an axe period. If it is to represent something else, we need a new graphic so we know what its true strength is.

[C Chulainn]

Here's an example of ridiculous luck in *my* favour against the AI. The situation was this: I'd invaded the Babylonians and been reasonably successful, but one of my Immortals was in serious trouble, having been cut off, reduced to one little red box, he'd been chased all the way across Babylonia to a coastal city and was surrounded. I thought "What the hell, nothing to lose!" and had him attack the city - defended by a fortified regular spearman, across a river. He won! I captured and razed the city! Of course, the next Babylonian slaughtered him, but still, it shows that fortune doesn't always favour the AI.