Compared to Ronald McDonald your military is Average!!

cracker · May 11, 2002, 19:04

Compared to So-in-So your military is Average!!

When the Sid Clone Military Advisor pops up and tells me this, I just want to shoot him with acid filled paintballs.

The CIV3 military assessment by raw body count is stupid beyond a normally forgivable game design issue because this one factor tends to drive extortion demands and stupid AI behavior that just should not be happening. If Psycho Chaka or Ballistic Bismarck thinks that our military is average this doubles our triples the chance that they will do something stupid to attack us.

The problem is that the CIV3 game designers applied an Ancient Age military assessment formula to every aspect of the game instead and included total unit body count in the military power formula. Based on test data, every unit is included in the formula. Workers, Settlers, Scouts, Explorers, Land Units, Air Units, Naval Units, Transports, and Bombardment Units are all counted equally in the military formula.

To test this process, I used the editor to artificially build treasury and population growth so I could build a force of 20 workers and one warrior to compare to the AI’s military force of 12 spearmen and archers and 2 workers. In this case, the AI doesn’t attack me and I get messages from the military advisory saying “Compared to these bozos our military is strong”. All other factors (cities, etc.) are equal. So what are my 20 workers going to do to defend the empire when those weak 12 military units come charging into the realm.

I tested the opposite case, and manipulated the units to have 12 Mech Infantry units and 2 workers in my military while the AI’s military contained 4 warriors and 18 workers. In this case, my wise and knowledgeable Sid Clone Military Advisor popped up saying “Compared to these bozos our military is weak”. The next turn when I asked two evil AI warriors to leave my territory, they declared war on me because they thought their military power gave them that right.

Everyone who really thinks this is a conscious and logical game design issue; hold up your hands now so we can identify you permanently as someone who has absolutely lost all contact with reality.

Inverse Icarus · May 11, 2002, 19:05

thats not sid. isnt that mike? sids the science guy.

cracker · May 11, 2002, 19:05

In several recent games, this became painfully obvious because I had crested a major technology lead over the 4 of the 5 AI’s that remained in a game, only to find the backward AIs would attack when I joined workers into cities to reduce to support cost of field workers after the entire continent had been railroaded. Since I was China and had the Industrious workers plus Democracy plus replaceable parts I was up to the point where two of our Chinese workers could clear pollution in one turn. I also had about 35 slave workers from early conquests of other civs. The slave workers did not appear in the military power count.

It’s just really not appropriate to have the military power ratio shift because I have Infantry while the other civs have spearmen and riflemen and just because I reduced the non combatant workforce that was no longer needed.

This cries out to be fixed in the next patch well before any further incorporation of this flawed logic gets passed on to multiplier of the XP version.

What is needed is a conversion to some sort of military power assessment that takes effect when the civs have progressed out of the Ancient Age. If there was hesitation to make this the default assessment for the military advisor, then it could be implemented by adding the calculation to the military advisor’s arsenal when an upgrade to each foreign embassy called the “Military Liaison” or “Military Attaché” was purchased.

Military Power assessment would use the ADM and BRF factors to calculate the offensive and defensive power contributions of any given unit.

I processed several test games for the military power calculations and produced examples from the May Tournament Game hosted on Poly by Bill Chin.

To determine the simple military power of a unit (simple being the case where unit experience is not considered as in conscript versus elite) I multiplied the A or D value of a unit time the number of that type of unit. Movement capacity was used to modify these strengths by multiplying each value by a factor that depended on the number of movement points in excess of the base value of 1. Movement was taken to have more of a positive impact on offensive power than it had on defensive power because in many cases defenders do not have the power of using their superior movement capacity to retreat or avoid contact with enemy.

I used a factor of 30% for the increase in offensive power that a unit with 2 movement points would have over a unit with 1 movement point but the same A and D values. An example of this comparison would be between an archer (2/1/1) and a horseman (2/1/2). For units with 3 or more movement points, I used a factor 60% for the increase in offensive power due to enhanced mobility.

Similar logic was applied to incorporating the impact of mobility into defensive power. The increase in defensive strength due to mobility is justified by considering the impacts of retreat and counterattack capabilities. I used factors of 20% and 35% for units with mobility of two and then three or more respectively.

The offensive military power of bombardment capable units was calculated by multiplying the bombardment strength times the rate of fire (number of potential hits) and then modifying this power up by 50% when the range was greater than 1 as well as modifying the bombardment power up by 20% whenever the unit had mobility of greater than 1 move per turn as in the case of most naval units.

The defensive military power of bombardment units was very restricted because defending bombard units only got one shot per attacker and had no defensive capability on their own. Basically this calculation was just the bombardment strength of the unit times the number of units of that type.

Once these military factors where calculated they could be combined to provide an overall military power assessment of the units that were in the military force for a civilization. The offensive bombard powers were combined with the attack strengths to give an overall assessment of the offensive power of the military. Bombardment power was combined into the total power calculation by multiplying the bombardment power by 25% to account for the fact that bombardment power must be combined with offensive and defensive units in order to achieve effective military strength.

The total military power of a civilization was taken to be the simple average of the offensive and defensive military powers even though we should acknowledge that it would truly be mire accurate to compare the offensive military power of one civilization to the defensive military power of other rival civilizations in order to derive a more effective basis for decision making.

The results of these test calculations yielded an effective assessment system that can be simply implemented to eliminate the current discrepancies where the CIV3 AI’s are programmed to consider 2 warriors to be a superior military force to just one infantry or rifleman.

Jon Shafer · May 11, 2002, 19:10

Yes, the game has many serious issues, we know. Hopefully they'll be fixed in the XP... if not, then I'll get back to working on my own game again.

But anyways, where in Colorado are you from?

Alfonso · May 11, 2002, 19:17

Thrawn05 · May 11, 2002, 19:41

Okay, I give up. Is this a troll or not?

Oerdin · May 11, 2002, 20:07

This is a serious game design flaw that causes the AI to act irrationally. Why do some people try to shut down the conversaion by calling it a troll?
It seems like Firaxis could fix this easily jut by writing a code that takes each unit, counts up it's offensive and defensive power and then sums the result for each nation. That way we wouldn't have countries with nothing but workers declaring war on super powers.

ChaotikVisions · May 11, 2002, 20:38

If they removed settlers/workers from being part of your military, it wouldn't be too bad then. Since some people even today think superior numbers will beat superior training/tech. It'd be nice if it went off the value of your units a/d, but for a quick fix(as in the next patch instead of several months away in the XP), they could just remove the settlers/workers. But yes, this is pretty annoying, but then again there attacks aren't too hard to fend off anyway.

Carver · May 11, 2002, 20:44

Plant a spy with all other civs, this is the only way for you to know anything. The AI's swordsmen will still wind up charging your infantry - but oh well.

Zurai001 · May 11, 2002, 21:32

Quote:

Originally posted by Oerdin
This is a serious game design flaw that causes the AI to act irrationally. Why do some people try to shut down the conversaion by calling it a troll?
It seems like Firaxis could fix this easily jut by writing a code that takes each unit, counts up it's offensive and defensive power and then sums the result for each nation. That way we wouldn't have countries with nothing but workers declaring war on super powers.

I believe he's talking about Alfonso, he of the posts-a-laughing-face-in-every-thread.

Akka · May 11, 2002, 21:47

It's even more strange that they choose to count the mere numbers of units, as in previous civs and in AC your power was calculated by adding the A/D values of your units.
So here is really a design decision I don't understand ???

notyoueither · May 11, 2002, 22:27

I don't know. I really think counting workers and settlers is silly. Agreed. But the formula they should use is...?

I don't think AD multiplicative is so good. Cav don't hold ground. I'd say add the higher of the AD ratings plus the 1/2 of the bombard values for 0 AD units. Something like that.

Although, I agree. Legions should count for 9.

BTW. Did you consider that counting workers makes it easier for builders to do their thing in peace? Maybe not, but that's the only reason I can think of to count workers.

cracker · May 11, 2002, 23:32

I completed some example tables so it would be easier for people to visualize how this military power assessment should be implemented.

Here is a military power assessments for the military of CHine being played by the human player in Bill Chin's May tournament game:

cracker · May 11, 2002, 23:33

And here is the militray power assessment for India being played by the AI player:

cracker · May 11, 2002, 23:38

This Table provides a side by side comparison of the two military forces (China vs India):

Jon Shafer · May 11, 2002, 23:47

Yes, that's a good idea. The way it is now is dumb. But is anything really going to change?

And where in Colorado are you from!

notyoueither · May 11, 2002, 23:48

cracker.

Do you think an Infantry unit would survive being attacked by 30 Horsemen?

Do you think 1 Cav would survive approaching 9 Spearmen?

cracker · May 11, 2002, 23:56

What is both interesting and a bit silly, is that the AI thinks that the military power difference between China and India is only 4% as shown by the total ratio of unit count near the bottom of the 2nd column in the comparison table.

When the non-combatant units (workers/settlers/scouts/explorers etc.) are excluded the ratio begins to more clearly reflect that India is not currently a match for the Chinese military in the ratio of combat unit count since China has a 13% advantage.

When we apply the power calculation we begin to see the relative power of Cavalry and Infantry in Chinese army versus the Riflemen, Spearmen, and Swordsman of the Indian army. Chine currently has 1013 offensive military power points versus only 530 offensive points for India and this indicates a 91% advantage in offensive power.

A more revealing comparison ratio is found in the very last row of the comparison table. This "A-vs-D Ratio" compares the Offensive Power of China to the Defensive Power of India and results in a total advantage assessment of 39% if China were to simultaneously attack every Indian military unit with all the Chinese Military units. If India were to attack China, the "A-vs-D Ratio" is reversed and shows that the defensive Chinese units have a 90% advantage over the Indian offensive power.

In reality, military success is almost always a matter of creating localized conditions where your attacking or defending units outpower the corresponding units of the enemy.

But at least the use of some sort of military power ratio would reduce the frequency of the silly advisor messages that under evaluate the military situations and also lead to the absolutely suicidal AI behaviors where we see attacks on modern age units by groups of outdated AI units that have not been upgraded.

cracker · May 12, 2002, 00:22

NeitherYouEither,

Your posted examples are a bit ridiculous and out of scale for the discussion here.

the 1 Infantryman versus 30 horseman would calculate a military defense power of 10 while the 30 attacking horsemen would result in a calculated offensive power of 78. So definately the the horsemen would have an overall advantage due to shear numbers. In the abstract sense, having and army of 30 horsemen would give you almost an 8 to 1 advantage over a civilization that had only 1 infantry unit in their army. The real point would be that the 8 to 1 advantage would be far more realistic than the 30 to 1 advantage you might assume if you just looked at the unit count.

In reality, a single infantryman fortified (+25%) in a city of population 7 (+50%) located on hills (+50%) could very well fend off all the attacks of an army of horsemen because the effective defensive strength of the infantryman would be 22.5 versus only 2 attack points for each of the horsemen.

I have in fact had a single infantryman survive an attack by 32 conscript barbarian horseman (but it was close).

Your other example of the cavalry unit attacking 9 spearmen defenders would depend very heavily on the terrain and fortification status of the spearmen. But using the Military power assessment would give us an effective offensiive power of 10 for a military force of 1 cavalry unit compared to a defensive power of 18 for a military force made up only of 9 spearmen. The ratio of about 2 to advantage for the defenders is again a much clearer representation of the realtive strengths instead of the tunnel vision approach of a 9 to 1 advantage you would calculate by just using the raw body count of units.

So my real answer is that I do not think that either of the two extreme examples that you cite would result in survival of the 1 units extremes.

Let me pose the question more in a way that would reflect how the AI makes decisions:

How many Jaguar warriors out of an attack force of 12 will survive an attack on a city defended by 4 rifleman even though you wise Military advisor has told you that your military is 3 times stronger than your enemy that has the riflemen?

Coracle · May 12, 2002, 01:02

Quote:

Originally posted by cracker
I have in fact had a single infantryman survive an attack by 32 conscript barbarian horseman (but it was close). . .

My God, what an example of TEDIUM.

Conclusion: I can't believe anyone is playing with the idiotic, non-historical unit values that Firaxis gave us. I Edited them in December.

miike · May 12, 2002, 01:28

Ok, so how can you edit how the AI adds up the military values. I think that everything should have a number. The stronger the unit, the higher the number. Then it is simple algebra to figure it out. Why didn't Firaxis see this sooner? This seems like a pretty big flaw if you have a bunch of people actually developing the game.

Todd Hawks · May 12, 2002, 01:53

On the original topic:

You all might want to know that Soren stated some time ago that the AI DOES take into account the strength of a unit.

Only the military advisor doesn't. So you get wrong readings on the relative strength of your military but when the AI has to decide whether to attack or not it doesn't use the military advisors count.

Th0mas · May 12, 2002, 02:57

*poof*

...and the thread dissapears in a puff of irrelevance

Skanky Burns · May 12, 2002, 04:49

Actually, no Th0mas.
As stated by cracker, the AI still does make bad decisions (like declaring war) based on comparative military strengths, when common sense would have them not. The function may take into account the strength of the units, but the evidence suggests that it is not properly balanced.

Kilroy_Alpha · May 12, 2002, 06:18

Quote:

Originally posted by Todd Hawks
On the original topic:

You all might want to know that Soren stated some time ago that the AI DOES take into account the strength of a unit.

Only the military advisor doesn't. So you get wrong readings on the relative strength of your military but when the AI has to decide whether to attack or not it doesn't use the military advisors count.

He was refering to an individual decision to attack a certain unit or not. There are several levels of AI decision making: Global, city, and unit specific. The unit-level AI takes relative strengths into account - the global AI does not. So the AI will change the way it fights depending on your actual strength, but it's tendency to make war is dependant on unit counts only.

notyoueither · May 12, 2002, 06:35

I believe Todd is correct. There would be 0 point in bringing up the MilAdvisor in unit-on-unit calculations. And I recall, since he reminded me, that the topic under discussion was the AI evaluation of relative strengths of Civs.

The Rook · May 12, 2002, 08:46

Todd may be correct, but the AI does go to war sometimes when it obviously shouldn't.

Todd Hawks · May 12, 2002, 10:20

Found the thread.

And Soren is clearly NOT talking about a decision to attack one unit.

http://apolyton.net/forums/showthrea...threadid=44074

Excerpt:

"The strength algorithm that the AI uses to determine power is [...]: It counts up military units and gives them a value according to their attack and defense values. This value is "Power" on the Historgraph (including some of factors including techs/cities/resources/wonders...)

The algorithm the foreign advisor uses to tell the user relative military strength is not used by the AI in any way. It is only used by the foreign advisor. (Also, it does not count workers. I'm not sure where that rumor got started... perhaps it was a bug in 1.07. At any rate, it does not count non-military units in 1.17. If you have experienced differently, please post a save.)"

Flanker · May 12, 2002, 11:07

Who cares, war is fun...

Hurricane · May 12, 2002, 12:29

Well, it sure seems in Coracle´s example that the mil. advisor sure enough counts workers, too. So I suggest, Coracle, that you email your findings to Soren. Maybe he will look at it.