combat odds display - a possible bug ?

[kmezo]

When I attack a city with more than 1 defender unit the combat odds calculator appears to be broken or misleading to me.
As an example I attacked a barbarian city on grassland defended by 2 archers. One of the defending archer had 2 promotions in city garrison, my swordman had a city attack promotion.
Still the odds were displayed like 23 % .
Once I killed the first archer and attacked with a fresh swordsman their second archer the odds display was realistic ( when only one defender was left )

I was thinking maybe the 23 % odds means that I have 23 % chance to kill _both_ defenders with the same _one_ attacker unit ... but even if this is the case that is quite a useless information.

[gilfan]

3 base archer
+50% city
+50% promotions
+25% fortify

that's 6.75 right there, ignoring any city culture bonuses. If your swordsman was wounded at all, if you were mistaken and the city was on a hill, if the city had any cultural defense bonuses, if you were attacking across a river... any of those could account for the 23% odds. I've never seen the sort of bug you think you experienced. I'm inclined to think you just overlooked something here.

[Son of David]

Yep, there are bugs in it. As I pointed out in my last thread, sometimes combat odds are displayed in excess of 100%.

Sometimes units will disappear from view when zooming in to attack, as well.

[Nacht]

Apparently the 100+ bug has to do with both units having first strike. And first strike is yet another bonus of the archer. I'm with gilfan in thinking that something has been overlooked. There are quite a lot factors involved.

[Blake]

Technically the AI's know (or can know) everything about the state of the game and the only question is what they use in the decision making processe. Or at least I don't THINK Civ4 uses any kind of isolation layer between the game engine and AI.

And the other question in each case is whether the AI is cheating, or the human is deprived of information the interface should make available. The most glaring case possibly being "you traded with our worst enemies".

[gilfan]

well, you could look up who the worst enemies of your allies are in the foreign advisor, but I agree that they could make it a lot easier. maybe the ability to designate other civs as "allies" and get a warning when a trade would make them mad, or maybe just a warning if it would anger anyone you have a certain minimum positive modifier with.

[kmezo]

Thank you for the answers. I may have made a mistake in the calculation then.

[mkorin]

Easy my to prove or disprove

An easy but tedious way to prove or disprove the combat odds is keep records of combats over a length of time.

IE

95 %
W
L

90%
W
L

83%
W
L

However, you will have to keep these records for a long time in order to give it statistical accuracy. Remember there is an 83% chance of not rolling a given number on a six sided die or conversely a 17% chance of rolling it. Saying that there is no memory for the die so you could get 3 or 4 of your number in a row and then not get it for 40 rolls to give you the proper odds. My education, training and inclination is in Statistics and my rough quess opinion is that they are pretty close to what the outcome really is. However, until someone makes a table and takes the time to prove it we can't know for sure. If there is some interest, I might take the time to proof it one way or the other.

Mike

[Solver]

The combat odds are always buggy if both the attacker and the defender have first strikes.