A look again at combat test

etj4Eagle

ACME Salesman
Joined
Dec 6, 2001
Messages
614
Location
Columbus, OH
This latest patch gives us the opportunity to look again at those infamous combat match ups and verify for ourselves that combat is really working as it should (at least for those who still believe the computer plays with loaded dice). The ability to have the seed reloaded and the ability to place starting units makes this possible.

In the barely functional state that I am after driving for 12 hours I decided to give it a whirl with a modern armour - Infantry and a marine - spearmen match up. There would have been a Tank warrior match up, but I forgot to set the starting number of enemy warriors to 25.

The two battles were on deserts so the only defense bonus was the basic 10% bonus.

For the MA-I matchup, the attacker won 72 rounds and the defender 33 for a win precentage of 68.6%, exactly matching the predicted value of 68.6%

For the Marine-Spearmen matchup, the attacker won 68 battles and the defender 26 battles. This gave a win percentage of 72.3%, a little shy of teh predicted 78.4%

While 199 rounds is still statistically insignificant, it does appear to be working correctly. And I believe that within a short period of time this "issue" can finally be put to rest forever.
 
Good lord are you telling me you reloaded the game that many times to test this out?? :eek: :eek:
 
Originally posted by Dominix
Good lord are you telling me you reloaded the game that many times to test this out?? :eek: :eek:

Nope :) I am not that nuts (at least yet anyways). I only loaded the scenario once. However I had placed 50 pending battles on the map. And as you notice the battles took on average 4 rounds to solve each (ie on average the attacker lost one 1hp).
 
After running another round of tests here are the current results:
A D Combat Rounds Actual win precentage Expected
24 10 207 69.6% 68.6%
8 2 94 72.3 78.4
16 2 103 86.4 87.9
8 6 135 54.8 54.8

Again the random number generator and combat appears to be working correctly. And this time I did include that infamous tank verse spearman matchup. And as you can see the defender is expected to win a round 12% of the time on flat terrain. Since combat might only 3 rounds long (regular units), a series of three bad die rolls there is very possible.

I see no indication that there is anything buggy about the combat system or random number generator. Though I still should test combat terrain with combat bonuses to verify the results there as well at some point.
 
Originally posted by etj4Eagle
I see no indication that there is anything buggy about the combat system or random number generator. Though I still should test combat terrain with combat bonuses to verify the results there as well at some point.

I can already hear the critics coming. "Your results prove nothing - it´s the lone spearman in the capital that never dies" or "Sea battles are more buggy" and so on. But it sure looks like the number of whiners is decreasing.
 
For the Marine-Spearmen matchup, the attacker won 68 battles and the defender 26 battles. This gave a win percentage of 72.3%, a little shy of teh predicted 78.4%



Wait a minute. Are you telling me that about a quarter of the time spearmen BEAT marines??

They should NEVER beat marines.

All my post-gunpowder units have been Edited way up in strength. The only reason they are lower in the game is to "give those civs missing a needed resource a chance". Firaxis admitted that on this forum. In otherwords, because they made their resource allocation so absurdly low, they had to make unit values even dumber.

Edit.
 
Originally posted by Zouave

Wait a minute. Are you telling me that about a quarter of the time spearmen BEAT marines??
No, he's telling that that about a quarter of the time spearmen took one HP from a marine. That means that spearmen beat marines 7% of the time. A lot less than 25%, but it will happen now and then. If both were veteran, the spearman will win less than 5% of the time.


They should NEVER beat marines.
Yes, they sometimes should!
I can imagine you playing Risk: "Your army consists of 25 units, but mine consists of 27 units, so your army should NEVER beat mine."
What fun!:crazyeye:

All my post-gunpowder units have been Edited way up in strength.
Good for you, but don't instist that your way is the only correct way of playing. There is actiually some people (including me) that prefer to play it the way it was designed: having at least some challenge even when you have a tech lead.

The only reason they are lower in the game is to "give those civs missing a needed resource a chance". Firaxis admitted that on this forum. In otherwords, because they made their resource allocation so absurdly low, they had to make unit values even dumber.
This is not true. I read the quote, and they did not say that the resource part was the only reason they made the combat system the way they did.

I guess it haven't occured to you yet, but there might be a tiny, tiny chance that some of the design decisions you don't like was made - not only because of stupidity - but maybe because of a different, but still valid point of view.
Think about it. :confused:
 
In my game, I'm the French with 2 musketeers (defense 3) and a spearman (defense 2) defending a city, and the AI is attacking with 8 Riflemen (attack 4). The first 2 riflemen took out the 2 musketeers, leaving 1 spearmen vs 6 riflemen. Was I worried? Hell no, I had a SPEARMAN for defense, the best Civ3 defensive unit. Sure enough, it withstood the 6 battles, loosing only 1 HP. The AI never stood a chance ;) Its a pity I upgraded the other 2 spearmen to musketmen, I should have left them as spearmen.

Firaxis feature request - please allow us to downgrade units, so that we may get better defensive units.

<PS - the last bit I'm being sarcastic>
 
Originally posted by Zouave




Wait a minute. Are you telling me that about a quarter of the time spearmen BEAT marines??

They should NEVER beat marines.

All my post-gunpowder units have been Edited way up in strength. The only reason they are lower in the game is to "give those civs missing a needed resource a chance". Firaxis admitted that on this forum. In otherwords, because they made their resource allocation so absurdly low, they had to make unit values even dumber.

Edit.

As TheNiceOne said, the combat round victory is NOT winning a combat engagement, but just one round. One round of combat is defined as one die roll which results in one of the units loosing one HP.

I am not dealing with whether or not the A/D values that Firaxis gave were good or not. I am just verifying that they do give the expected combat results. If you wish to debate that, take those concerns to one of the other threads that deal with that. This thread is just to deal with probability and the random number generator portion of combat.
 
Originally posted by etj4Eagle
I am not dealing with whether or not the A/D values that Firaxis gave were good or not. I am just verifying that they do give the expected combat results. If you wish to debate that, take those concerns to one of the other threads that deal with that. This thread is just to deal with probability and the random number generator portion of combat.

BTW, thanks for the work. I've never believed the people that complain about how civ3 cheats with battle results. And it's good to see that it doesn't cheat (here).
:)
 
Since I saw some more posts about the super spearmen, I thought I would do a few more tests:

Tank V Spearman in Jungle and over River
16A V 2D w/ 25%+15% bonus
for 181 combat rounds found a 85.1% win verse the predicted 85.1% value

Mech Inf V Spearman on desert over River
12A V 2D w/ 10% + 15% bonus
for 172 combat rounds found a 83.1% win verse the predicted value of 82.8%

The spearmen perform exactly as expected, with no special bonus as some seem to believe. Also notice that the defense bonus for a river is only 15%, I believe it is given as 25% elsewhere in the reference part of this site.
 
Originally posted by smallstepforman
Firaxis feature request - please allow us to downgrade units, so that we may get better defensive units.


I always defend with spearmen now. You should see the AI's face when 20 of his advanced armor are destroyed by my lone elite spearman.

;)
 
Back
Top Bottom