MW Armies

The civilopedia has all of defense bonuses in it. Flat is 10%, hills 50%, mountains 100%, etc. There is a thread around here about the combat calculator that will tell you everything important, and you can use the calculator.

I am pretty sure it just adds one after another, they don't add 50% on top of a boosted 25%. So a spear fortified, in a town, behind a wall, on grassland would have 2+85%=3.7 defense. With with a wall or hills or city, the spear beats the sword. Without it, the sword wins. This is actually one of the nice things I like about civ3, the prepared defender can do well, but the unprepared defender has the disadvantage.

Also, a few artillery units will rapidly decrease the amount of units one loses, this is my main method of getting around the RNG.
 
When the average available attacking unit has an attack of 3 and the average defensive unit has a defence of 2, what is the exact defensive value of:

There are no towns and cities on jungle or forest tiles. When you found a town, they are automatically cleared. That leaves you with either a 10% or a 50% defensive bonus for terrain for towns and cities.

Before the advent of the Civil Defence building, what other defensive bonuses are available if I've missed any?

If the attacker has to cross a river that adds 25% to the defender. And then there are radar towers, also 25%, right?


I have excluded Worker built Fortresses as I've never seen the AI build one as yet and I rarely do (though when I do they do deter the AI superbly, I just rarely get good locations for them).

The AI loves to build those on chokepoints.
 
Just last night, full health 13/13 sword army died redlining a 5/5 spear fortified in a city size 2, no wall, on grassland (no river). :mad:
 
Thanks for that excellent link.

Here's some easy on the eye statistics I've calculated using creamcheese's accumulative method of defence bonus stacking:


Criteria:

For the Ancient Age maximum defence bonus is as follows: Fortified 25%, Walls 50%, Hill 50%, other side of a River 25% = 150% maximum defence bonus.

For the Ancient Age normal defence bonus is as follows: Fortified 25%, Grassland 10% = 35% normal defence bonus

----------

For the Medieval Age maximum defence bonus is as follows: Fortified 25%, City 50%, Hill 50%, other side of a River 25% = 150% maximum defence bonus.

For the Medieval Age normal defence bonus is as follows: Fortified 25%, City 50%, Grassland 10% = 85% normal defence bonus.

----------

For the Industrial Age maximum defence bonus is as follows: Fortified 25%, Metropolis 100%, Civil Defence 50%, Hill 50%, other side of a River 25% = 250% maximum defence bonus.

For the Industrial Age normal defence bonus is as follows: Fortified 25%, Metropolis 100%, Grassland 10% = 135% normal defence bonus.

----------

For the Modern Age maximum defence bonus is as follows: Fortified 25%, Metropolis 100%, Civil Defence 50%, Radar Tower 25%, Hill 50%, other side of a River 25% = 275% maximum defence bonus.

For the Modern Age normal defence bonus is as follows: Fortified 25%, Metropolis 100%, Grassland 10% = 135% normal defence bonus.

----------

I have not bothered to calculate Warriors, Aztec Jaguar Warriors, Horsemen, Iroquois Mounted Horsemen, Archers, Chariots, Egyptian War Chariots, Scouts, Incan Chasqui Scouts, Longbowmen or Explorers. Scouts and Explorers have a defence of zero and all the others have a defence of 1. The most a defence of 1 can ever get before the Industrial age is 2.5, so obsolete from the get-go really.

I have not calculated anything but land troops versus land troops.

I am assuming you are attacking the Capital City or decent second or third cities.

----------

Spearmen, Zulu Impi, Sumerian Enkidu, Swordsmen, Celtic Gallic Swordsmen, Persian Immortals, Ancient Cavalry, Babylonian Bowmen, Mayan Javelin Throwers, Hittite Chariots, Medieval Infantry, Viking Berserkers, Arab Ansar Warriors, Mongol Keshiks, Spanish Conquistadors all have a base defence of 2 providing the following:

Ancient Age Max defence of 5
Ancient Age Normal defence of 2.7
Medieval max of 5
Medieval norm of 3.7
Industrial max of 7
Industrial norm of 4.7
Modern max of 7.5
Modern norm of 4.7

So even under normal conditions a Swordsman or Medieval Infantry or Knight only just beats a Spearman and the Spearman can quite quickly and easily surpass them. Another reason why Cavalry really speeds up the conquesting. A max of 7.5 makes defending against an attack of a tank's 16 seem kind of impossible. More of that later.

Pikemen, Greek Hoplites, Carthaginian Numidian Mercenaries, Roman Legions, Crusaders, Knights, Chinese Riders, Indian War Elephants, Cavalry, Russian Cossaks, Ottoman Sipahi all have a base defence of 3 providing the following:

Medieval max of 7.5
Medieval norm of 5.55
Industrial max of 10.5
Industrial norm of 7.05
Modern max of 11.25
Modern norm of 7.05

So even under normal conditions a Cavalry, Infantry or geurilla only just beats a Pikeman and a Pikeman can quite quickly and easily surpass them. Another reason why Cavalry really speeds up the conquesting still. A max of 11.25 makes defending against an attack of a tank's 16 seem more plausible, but not really. More of that later.

Musketeers, Dutch Swiss Mercenaries, Japanese Samurai all have a base defence of 4 providing the following:

Medieval max of 10
Medieval norm of 7.4
Industrial max of 14
Industrial norm of 9.4
Modern max of 15
Modern norm of 9.4

So even under normal conditions a Cavalry, Infantry or geurilla should be desperately struggling against them. This should slow down conquesting until Tanks. A max of 15 makes defending against an attack of a tank's 16 seem more plausible. More of that later.

French Musketeers have a base defence value of 5 providing the following:

Medieval max of 12.5
Medieval norm of 9.25
Industrial max of 17.5
Industrial norm of 11.75
Modern max of 18.75
Modern norm of 11.75

So even under normal conditions a Cavalry, Infantry or geurilla should be losing regularly against them. This should slow down conquesting until Tanks to which the tank would then be only a slight advantage. A max of 18.75 makes defending against an attack of a tank's 16 seem totally plausible. More of that later.

Riflemen have a base defence value of 6 providing the following:

Medieval max of 15
Medieval norm of 11.1
Industrial max of 21
Industrial norm of 14.1
Modern max of 22.5
Modern norm of 14.1

So even under normal conditions a Cavalry, Infantry or geurilla should be totally losing against them. This should slow down conquesting until Tanks to which the tank would then be only a slight advantage. A max of 22.5 makes defending against an attack of a tank's 16 seem quite safe. More of that later.

Infantry have a base defence of 10 providing the following:

Medieval max of 25
Medieval norm of 18.5
Industrial max of 35
Industrial norm of 23.5
Modern max of 37.5
Modern norm of 23.5

So even under normal conditions attacking them with Cavalry, Infantry or geurilla should be suicide. This should slow down conquesting until Tanks to which the tank would then still be at quite a disadvantage. A max of 37.5 suggest even Modern Armour should not bother attacking them. More of that later.

TOW Infantry have a base defence of 14 providing the following:

Modern max of 52.5
Modern norm of 32.9

So even under normal conditions attacking them with any land unit should be suicide or at the very least a hard fought struggle for Modern Armour. More of that later.

Mechanised Infantry has a base defence of 18 providing the following:

Modern max of 67.5
Modern norm of 42.3

So even under normal conditions attacking them with any land unit should be suicide. More of that in the following analysis.


Conclusion

I suspect when they beta tested the game they found that either the defence stacking wasn't working and just didn't tell anyone or they added a secret bonus to all attackers. I know when I first played civ3 I was stunned that my Infantry could drop like flies against Cavalry, as if I might as well have left the city defended by Spearmen. This was always a bigger shock to me than having a Spearman defend a tank once every few games.

This makes me think that the numbers aren't as important as they appear to be at face value.

At the start of the game a Warrior 1/1/1 can attack and kill a Spearman fortified in a town 1/2.7/1. That's almost a 300% assault. Even out in the open defeating a 1/2/1 Spearman or Archer is a 200% assault.

By viewing the numbers in terms of percentages rather than numbers one can then rationalise how a Cavalry 6/3/3 can assault an Infantryman 6/18.5/1 because this is, again, just under a 300% assault. Similarly, a Cavalry assaulting a Mechanised Infantry 12/18/2 out in the open is just a 300% assault.

Another two options to consider are, firstly, that the Random Number Generator has a much wider range than is obvious and that this generator changes during each era. So in the Ancient Age all the rolls might be calculated on the bases of a 6 sided dice to which an attack of 1 attacking a defence of 3 gives the attacker a win potential on a roll of 4, 5 or 6 where the defender rolls a 1, 2 or 3 respectively. Moving onto the Medieval era might raise the dice value to a ten sided dice, Industrial 20 sided and modern 50 sided, so a Modern Armour 24/16/3 could roll up to a 74 to beat the Mech's normal 12/42.3/2 when rolling under 20.

On top of this the game might have critical hits, similar to roll playing games, where a maximum roll is an automatic hit and a minimum roll is an automatic loss. This would occur much more frequently at the six-sided stage and very rarely at the 50 sided stage.

So, how can a Tank lose to a Spearmen? In the modern era, applying a 50 sided dice, a Spearman 1/4.7/1 could roll a 30 to score 34.7 whereas the Tank 16/8/2 could roll an 8 to give an attack of just 24. Now it makes some sense. So, for a Veteran Tank, with 4 hit points it is likely to lose to a critical, get a minimum critical and then get two low scores matched with two high scores for the Spearman and die on the rare, but not totally infrequent, occasion.

So, perhaps your Army, attacking at 3/1/2 to a defence of 1/2.7+/1 simply met with 12 terrible rolls, something which doesn't happen often but seems to happen more noticeably with computer generated Random Number Generators. I remember playing The Temple of Elemental Evil and having exactly that kind of luck with my Archer throughout most of the game.

If anyone knows the exact stats behind encounters I'd be glad to be thrown out the ball-park. The odds calculator is good, but what is it's formula and where is the formula described in full so I can stop making myself seem like a fool newb with random guesswork?
 
The army was not 3/1/2 unit at that point; I believe it is something like a 6/4/2 unit, though the math has always been a little odd to me. But I do get your point. I have made several sacrifices to the RNGods in the hope that they will forgive my trespasses.

Percentages being what they are, they do miss certain miscellaneous factors as to how the battle will result, such as: Current hit points of both units, retreat capability (or prevention), and the defensive shot - both from archer-types and artillery on the defense.

After the pre-calculated attack results generated by Civ4Col I have come to decide that these calculations to do have much bearing on the actual outcome of battle. And if the variance is that large, then precise prediction is probably impossible; such may be the case and the intention. And, of course, Murphy will always make an appearance. The rule clearly states that if you are using your last unit vs. the last redlined defender you will flawlessly lose 99% of the time. :D
 
The basic formula is supposed to be that you roll a random number function of (a+d) 'sides', where a = attacker's attack value, modified by all possible bonuses (there are no bonuses for attackers), and d = defender's defense value, modified by all possible bonuses, which we have already discussed extensively. And if you get a value 1 - a, the attacker wins, otherwise the defender wins.

e.g. a 1hp sword attacking 1hp spear on a terrain with no defensive bonus (there is no such terrain) has a 3/5 chance of winning.

How do we know this? Probably because the developers told us. Does that make it true, or definitive? No. Now, community members, who I'd trust to the know the game better than the devs any day of the week, have certainly run extensive tests of the mechanics and probabilities, using editor-created testbed scenarios. While these should have verified the basics, it is certainly possible that there are extra modifiers hanging around (e.g. do units get a combat bonus from adjacent friendlies, or a combat penalty for attacking with a yellow movement indicator?) that weren't tested for.
 
Well... it just happened to me this very second.

Playing a Tiny Emperor Pangea. My advantage is producing Swordsmen at the drop of a hat and having a tech lead even if falling behind on number of towns.

Anyway, my first assault on the enemy Capital City is a resounding success. Occupied and reduced to size 1, Oracle captured. This battle generates a leader who races back to base camp. In the mean time my Swordsmen are overwhelmed by non-stop invading forces (the difference between Monarch and Emperor here is insane).

I then hold back a small onslaught from 2 civs while I found my 4th town in order to build an army. The army then proceeds to hack my area clear and re-march on the AI's Capital City, which it does superbly.

One more step to go and I've pretty much won the Ancient Age and set myself up well for the Medieval phase. The AI's Iron deposit is in a town just 5 squares from the Capital I'm occupying. So my army restores it's health and goes on the march.

1st Spearman dies easy. Wait for next turn. 2nd Spearman dies easy, so far, in all, maybe 3 Hit Points lost. 3rd Spearman (probably the last) doesn't lose a single Hit Point and my army just stands there losing Hit Points then dies. Just like that.

Firstly - this goes against all forms of expectation. How is one supposed to make rational decisions about anything if you can't even rely on something foolproof.

Secondly - Does this only ever happen at crucial junctures? I know all battles are crucial, but this particular moment was so completely the turning point, the point between me having Iron and my neighbour not in the Ancient Age approaching Medieval, that it's a complete game killer. There is no point moving forward from here. It's as if the game is playing some cruel joke when in reality it's just some kind of F'ed-up mechanics.

*shrugs*

The whole game felt ultra contrived at almost every step from a very early stage, even more so than I'm used to seeing.

Maybe this is why I don't like the higher levels, too much hinges on the outcome of irrationality rather than the lower tier of levels where you can goof up quite a lot and bounce back quite easily. It must be heartbreaking to get close on the really high levels and for something like this to happen.


you roll a random number function of (a+d) 'sides', where a = attacker's attack value, modified by all possible bonuses (there are no bonuses for attackers), and d = defender's defense value, modified by all possible bonuses, which we have already discussed extensively. And if you get a value 1 - a, the attacker wins, otherwise the defender wins.

e.g. a 1hp sword attacking 1hp spear on a terrain with no defensive bonus (there is no such terrain) has a 3/5 chance of winning.

How do we know this? Probably because the developers told us.

This makes no sense to me, even though it's written so well. It's still a cloud of dust unless you know what number range is being rolled, and there must be a varied number roll otherwise there be no winning or losing Hit Point, it would be all one-way traffic. Without knowing the roll range how can someone lay odds? Are you suggesting they are suggesting it's either a zero or a 1 roll?
 
Secondly - Does this only ever happen at crucial junctures?

Sometimes it sure can seem that way!:lol:

Maybe this is why I don't like the higher levels, too much hinges on the outcome of irrationality rather than the lower tier of levels where you can goof up quite a lot and bounce back quite easily. It must be heartbreaking to get close on the really high levels and for something like this to happen.

This is why artillery is so important on the higher levels, it reduces the chance factor and cuts down on losses. It is also probably the reason people can consistently win on Sid.
 
This is why artillery is so important on the higher levels

In this instance I had no need for them, everything was going swimmingly until the army loss. Also, mathematics had only been discovered about 5 turns ago.

Start with Pottery and Masonry, learn Bronze Working and Iron Working, trade for alphabet, learn writing, Code of Laws then Philosophy, get Republic free, trade for Warrior Code and Ceremonial burial and Mysticism, learn Literature, trade for Map Making, then finally get to Mathematics.

The army was outside the town with Iron at the crucial moment a good 10 turns before Mathematics was learned (and no-one else was learning it to trade).

Not to mention the fact that my economy was at it's complete limit with 2 defenders in each town, 4 Swordsmen 3 Workers and a settler. If I'm going to shell out on another unit, it's not going to be one catapult to then take 2/3 turns to build and a further 10 turns to go and meet the army at the gates of a town it should just walk into, I suspect the town may be better defended by then. Not to mention the fact that it will need an escort for the journey. Shall I send my army back to pick it up? All, of course, for a *possible* damage of 1 Hit Point to a defending Spearman.

I know just mentioning bombarding is the key to higher level advice, but please, not every scenario demands them and I have completed the same map from Chieftan to Monarch with consummate ease without them and looked like doing the same in this game but with greater early game pressure and I was still making a few silly errors.

There was no crucial tactical error made, in fact quite the opposite. The only crucial factor was an army dying to a Spearman, the whole point of this thread.
 
I know that people say that all this stuff is about defense bonuses and all that sort of thing but as far as I am concerned absolutely nothing matters or affects the outcome except The roll of a die on that exact turn. Nothing else matters; nothing.
 
I wasn't mentioning your game in particular, and no criticism was implied. I was just commenting that artillery can make a lot of randomness not matter.

Like the time I experimented with Trebuchents (sp?) and catapults I had won in battle. That the three Trebs all miss and the catapult hit.
 
Like the time I experimented with Trebuchents (sp?) and catapults I had won in battle. That the three Trebs all miss and the catapult hit.

You've missed the point, with a fair stack of artillery, you can attack the AI (obviously it doesn't work all the time, that's why you need a stack) without them being able to defend or kill your artillery.
 
You've missed the point, with a fair stack of artillery, you can attack the AI (obviously it doesn't work all the time, that's why you need a stack) without them being able to defend or kill your artillery.

I'm sorry, but I wouldn't recommend Catapults for any Ancient Age encounters. If you mean Artillery as in the unit entitled Artillery, then that's a different story, that's an Industrial Age unit based on a railed-up world and a far cry from lugging round Catapults with limited budgets and limited production.
 
I'm sorry, but I wouldn't recommend Catapults for any Ancient Age encounters. If you mean Artillery as in the unit entitled Artillery, then that's a different story, that's an Industrial Age unit based on a railed-up world and a far cry from lugging round Catapults with limited budgets and limited production.

Yes, limited budgets and limited production that can be used more efficiently by making arty (yes, I mean all bombard land units, not just the unit artillery). If you are losing large portions of your army every time you attack an enemy, you have to rebuild them, costing you production. If you are playing a game where you can just outproduce the AI and win that way, you should move up a level perhaps.
 
(there are no bonuses for attackers)

I thought radar towers gave bonuses to their own side on the attack, as well as on the defense rolls?

Buttercup

That's a great job done on the unit offense and defense comparisons. Very useful info.

My own suspicion is the game uses a percentage "roll" to determine combat, but that the roll is sometimes modified by something else. Either a "bonus" is applied to the role before it occurs, or is tacked on after. And this is where the "invincible AI unit syndrome" ;) comes from. Something got triggered that boosted the AI unit to near invulnerable status. The modifying programing could be event triggered, or triggered at random. It could be that events trigger a chance the program will add its boost. There are several ways I can think of how this program might work.

A straight addition to the AI in the combat roll.
A unit will not die type of thing that negates negative results.

What ever it is, it seems to be unit specific, with one unit being invulnerable, not usually any of the others in a stack as far as I can remember. It could be it can only happen to one unit per player or AI turn. Both a hp boost and throwing out negative combat roll results might also explain the over abundance of hp losses winning player units receive. With a percentage boost, sometimes the addition of that extra percentage to the roll is enough to kill the unit, other times, it adds enough to cause heavy hp loss. It would work something like this for a combat between equally matched units each with 3 hp and with no bonuses applied:

to 12.5% player wins, no hp loss
+12.5- 31.25% player wins, 1 hp loss
+31.25-50% player wins, 2 hp loss
+50-68.75% AI wins, 2 hp loss
+68.75-87.5% AI wins, 1 hp loss
+87.5-100% AI wins, no hp loss

Now say the AI boost is 25%. There will be no player wins without hp loss and there would be just a 6.25% chance of the player winning and losing 1 hp. The chance of a 2 hp loss for a winning player unit would remain at 18.75%. For the AI, there would still be the same 18.75% chance of either losing 1 hp or 2 hp, but the chance of losing none would go from 12.5% to 37.5%. If the AI boost is 50%, or more, and the odds are such they favor the player more than the AI, such a percentage boost applied to an AI unit's rolls would look like those combats where an AI unit was seemingly invincible, even though it was much weaker, and even had only 1 hp. If the unit got the boost for the whole turn, it might explain how some weaker AI units can absorb several attacks from much stronger units. It could also be that such boost is variable, the amount determined by chance and difficulty level of the game, for example.

An AI unit does not die type of program could simply ignore rolls where the AI unit looses, or those rolls the player unit got off lightly.

This is all just idle speculation of what could be the reason why the AI seems to frequently have unusual combat results and decent sized streaks of these, while my units get these unusual results much less frequently and less extreme examples of these occurrences of very good luck.
 
This makes no sense to me, even though it's written so well. It's still a cloud of dust unless you know what number range is being rolled, and there must be a varied number roll otherwise there be no winning or losing Hit Point, it would be all one-way traffic. Without knowing the roll range how can someone lay odds? Are you suggesting they are suggesting it's either a zero or a 1 roll?

The random number being generated is a decimal in the range [0,a+d). The result is used to inform a binary outcome: (0) attacker wins, or (1) defender wins. This process is applied one hit point at a time.


scratchthepich said:
I thought radar towers gave bonuses to their own side on the attack, as well as on the defense rolls?
Forgot about them. Indeed, the 'Pedia does (in an uncharacteristic fit of specificity) say that they give a +25% bonus to both friendly attack and defense, and that bonus doesn't stack for having overlapping towers.
 
The random number being generated is a decimal in the range [0,a+d). The result is used to inform a binary outcome: (0) attacker wins, or (1) defender wins. This process is applied one hit point at a time.

If my memory is right, in Civ2 the rng roll loss accounted for 1/10 of a hp and the game kept rolling till one side lost all its hp. Which for a 3 hp unit, would mean at least 30 rng roll losses. It was my impression this system was kept for Civ3?

Forgot about them. Indeed, the 'Pedia does (in an uncharacteristic fit of specificity) say that they give a +25% bonus to both friendly attack and defense, and that bonus doesn't stack for having overlapping towers.

Thanks. I read that in the civopedia also, but I don't trust what that says and prefer having what it says confirmed by experienced players, such as yourself. I have little experience with radar towers as the only time I remember seeing them in a stock game was way back when I first got the game and played a WW2 Pacific scenario. That was also my first, and probably the only time, I used armies in a stock game. :lol:
 
Not to mention the fact that my economy was at it's complete limit with 2 defenders in each town, 4 Swordsmen 3 Workers and a settler. If I'm going to shell out on another unit, it's not going to be one catapult to then take 2/3 turns to build and a further 10 turns to go and meet the army at the gates of a town it should just walk into, I suspect the town may be better defended by then. Not to mention the fact that it will need an escort for the journey. Shall I send my army back to pick it up? All, of course, for a *possible* damage of 1 Hit Point to a defending Spearman.

I'm not trying to be picky, but since we've gone around this carousel of fun several times, I think this might be an insight into some of your difficulties. As I don't know the full layout of your game, I'm not being overly critical, just bringing this up as a point of discussion.

Unless I am pressed on all sides by the enemy, only a handful of towns have defenders. If you are in Monarchy then perhaps you actually meant MPs, in which case I see your point, though unit costs should be less of a problem with Monarchy and only cities that need MPs will receive them as a means of holding down use of the luxury slider.

I once read a piece of advice that said every city should have two of the top-of-the-line defenders at all times. The AI would agree with you. But it will kill your game play and is in the top ten worst habits, IMO. In fact, if you have not even learned Math yet I would question why you have any defenders at all. I don't even bother with more than a token force of defenders until infantry. Two per city is way too much, even with raging barbarians running around.

I would suggest as a means of making a catapult rush looking more attractive, the 20 shields per spear and unit support you are using for those two defenders per city would be better put to use to make 20 shield catapults. The up side is that the cats will upgrade and you can use them on the attack all the time (well, as fast as they can move). Your two defenders per city only do you good if they are constantly being attacked and/or are covering more valuable combat units. As it is unlikely you are being attacked at every city regularly, these units are as value adding as a coastal fortress. And the cost of upgrading or replacing units that you are not even using is just throwing good gold after bad.

In my case the 13/13 sword army dying to the 5/5 spear was a calculated risk. In the game in question I did not make artillery until RP because the AI was completely outclassed. The loss of the army, while aggrivating, was not the end of my advance. I did have to scramble for 2-3 turns to reorganize, though. :blush:

I do agree with one thing though - it was a critical time in the game. I was nearly overrun by the AI on another front because I had to draw off resources to nail the coffin shut on the other war front. But that just adds to the fun of the game! :lol:
 
Back
Top Bottom