Great Leaders: Everything you always wanted to know

[sumthinelse]

Originally posted by thefrenchzulu
I find the smaller the odds of success, the better the change of getting a leader.

When I play the Persians, running around with immortals, I get my first leaders with a few archers!

I also find certain events and buildings increase your change of producing a leader.

When desperate in the early game, I start my FB! That normally results in one.

There more strats, by they are mine...

I know that you are trying to help and you believe that these affect leader creation. However, a very competent and reliable Firaxis programmer has given us a summary of the factors that really do affect leader creation, and he did not mention any of these.

I once had a theory that attacking a Zulu Impi could never create a leader, but I was wrong about that.

Here is the classic cause-and-effect fallacy: You look out he window and the sidewalk is wet. You conclude that it must have rained. However, the real cause was that somebody turned on sprinklers to water the grass. It's easy to jump to conclusions about what causes what. I know I have done it.

 

[Qitai]

This just comes into my mind. Sumthinelse> You may want to include the promotion probabilities here since it is very much related I think.

 

[thefrenchzulu]

Getting many early leaders will swing a game completely towards a human player.

Your very reliable Fixaris programmer of course is going to give you the secret to leader creation.

I have play tested leader creation during different ages. It is really not just 1/16 chance. If you honestly believe it is, tough on you.


Ever wondered why you get multiple leader per turn when attacking with MA, especially on their third attack? Is that still just 1/12 chance? I don't think so.

Another example:

If you attack with a stack of swordsmen, one in the group might be promoted. If you should count which one, and let say it the fourth one. IF YOU should reload and the fourth one is now elite, chance are you now have a leader. Fine that's cheating, but prove that elite promotion=leader.

Promotion is easier to achieve when the battle is against you, e.g. you attack invantry with immortal. This might sound stupid, but I often do this for leaders. (Take on one hit point invantry and do lose some immortals. You do win after 2/3 attept. So much for the great battle system)

I also use losts of old units to pull AI counter cavalry attacks. They love to rather go for my elite swordsmen on the hill rather than my riffleman stack in the open. More leaders.


Bottom line there are more to it than just plain maths. I find some key building play a role, but you can take my word on it or read some strats I've posted on ALL CIV's forum...

 

[sumthinelse]

Originally posted by thefrenchzulu


Ever wondered why you get multiple leader per turn when attacking with MA, especially on their third attack? Is that still just 1/12 chance? I don't think so.

Since my Thinkpad is down, I'm writing this on my old notebook computer, not big enough to play civ3. I will test this when I get my new computer next week. I will keep an open mind, but since you did not tell me how many times you tried this and how many times you created leaders, I don't know what to think yet.

Another example:

If you attack with a stack of swordsmen, one in the group might be promoted. If you should count which one, and let say it the fourth one. IF YOU should reload and the fourth one is now elite, chance are you now have a leader. Fine that's cheating, but prove that elite promotion=leader.

I'm not sure what you mean. Which unit in the stack creates the leader?

Promotion is easier to achieve when the battle is against you, e.g. you attack invantry with immortal. This might sound stupid, but I often do this for leaders. (Take on one hit point invantry and do lose some immortals. You do win after 2/3 attept. So much for the great battle system)

Again, you don't say how many times you tried this and how many leaders you got.

I also use losts of old units to pull AI counter cavalry attacks. They love to rather go for my elite swordsmen on the hill rather than my riffleman stack in the open. More leaders.

You are saying that when cavalry attacks your swordsman it is more likely to produce a leader than cavalry attacking a rifleman?
What percent of the time do you get leaders from the swordsman being attacked, and how many times have you tried this? What if another unit attacks your swordsman?

 

[sumthinelse]

Originally posted by Qitai
This just comes into my mine. Sumthinelse> You may want to include the promotion probabilities here since it is very much related I think.

Good idea Qitia. I changed the main section to include promotion probabilities.

 

[thefrenchzulu]

First my appologies for copy/pasting your thread. I'm new on this forum and haven't had the time to figuring out the quote/reply to functions.

I will keep an open mind, but since you did not tell me how many times you tried this and how many times you created leaders, I don't know what to think yet.

Sorry, I don't always keep detail stats like other players. An elite MA attacking and winning it's third battle per turn will nearly always result in a leader. I normally create armies immediately cause there are no wonders to rush at this stage. Have had three leader in one turn. (Two immediate armies) Have had numerous MA going from veteran to elite to leader creator in one attack turn.

If you attack with a stack of swordsmen, one in the group might be promoted. If you should count which one, and let say it the fourth one. IF YOU should reload and the fourth one is now elite, chance are you now have a leader. Fine that's cheating, but prove that elite promotion=leader


I'm not sure what you mean. Which unit in the stack creates the leader?

Will try to explain. If you attack a city with mulitple defenders, you attack with veteran units first. Let's say the fourth one gets promoted to elite.(It doesn't matter if you reload, the fourth one will always become elite.)

If reload and attack the fourth time with an elite unit (same as the previous units), you will get a leader. I have tried this a few times with 100% succes rate. Though this is cheating, I was actually play testing the militaristic attribute. It suppose to result in easier promotions, but I never get more leaders playing them??? Nothing else must be changed during the attack sequence, eg if you already have a leader and use it, you will get different results.


Promotion is easier to achieve when the battle is against you, e.g. you attack invantry with immortal. This might sound stupid, but I often do this for leaders. (Take on one hit point invantry and do lose some immortals. You do win after 2/3 attept. So much for the great battle system)

Again, you don't say how many times you tried this and how many leaders you got.

The invantry vs elite immortal, I normally get a leader from a succesful attack. I have created upto a hundred immortals in a game, so this is just a way of getting rid of them. Unfortunately I have no exact stats on this one, lets say 50% success.

You are saying that when cavalry attacks your swordsman it is more likely to produce a leader than cavalry attacking a rifleman?

I don't normally use this strat for leaders, but rather as a offensive strat. If you attack with cavalry, you don't want to lose to many to counter strikes. Moving obselete units in behind your cavalry, normally pulls the AI's counter cavalry attack. I do lose many of these units, but it pulls the AI cavalry into the open.


OK, these two are quite close and is all about unit disparity.

I want to stress that the bigger the unit disparity, the less your change on promotion. Less promotion, less leaders. That's why I do attack with archers early and not only immortals, unless of course I'm attacking the Greek.

This last part of unit disparity, have not been posted anywhere and it is difficult to put stats down.

I can say that I often play as the Persians and my Deity strat is 100% attack, using 30-40 early immortals! I have gone through games only getting my first leaders once I've killed musketmen. So I changed my strat and combine some archers into my offensive stacks. This results in many more early leaders!

If I find time one day, I'll actualy test this using the editor, like B.Speedy does. Should be an interesting posting...

You said nothing about my theory about certain building eg your FB increasing your chances. Must say this one is really difficult to test, but I have used it often for that first leader. (Might have posted this on a different forum?) Now this have not been discussed any where, but I do use it often. Sure the programmer of Firaxis didn't mention this one!


If you have a look at my GOTM10 posting, you will see that I had 4 leaders by 340AD. What I have not added is that I spend lots of money updating my World Map, once I've got it. Reason behind this is that it is cheap and the AI can see babarian huts. I used them for upgrading my horsemen so that I had numerous elite ones for attack. Think of it as training.
If I attacked with knights, rather than horsemen, I would have had easier battles, but less leaders!
(4 early leaders really wins you the game)

 

[Zachriel]

Originally posted by Lt. 'Killer' M.
To clarify:

I *think* that if I start Civ3 fresh and begin a new game, the first leader will take longer.
If I play for a while, then start a new game, it will appear at the normal ratios mentioned by sumthinelse.

Doubtful, though certainly possible. Here is an example of a Great Leader during the first major battle in the Ancient Age, Napoleon in 2710BC. Surprised the heck out of me!

Mon cher Napoléon
** SPOILER ** GOTM10 **
http://forums.civfanatics.com/showthread.php?s=&threadid=30172

 

[sumthinelse]

Killer said that he started CIV3.exe and immediately after that started a new game. Killer suspected that in such a game the odds of creating a leader would be lower. Why that would affect leaders, I would not know. This has not been proved or disproved.

 

[thefrenchzulu]

I have never tried the new game thing to generated leaders, but have a comment on generating your first leader with an elite warrior. Considering that the jaquar warrior was fortified in a city, gave it various defense bonuses. The probability of winning the battle was thus against you, increasing your change of generating a leader. My unit/battle disparity theory...

 

[Ironikinit]

Great posts, sumthinelse, thanks for writing them.

frenchzulu, I once had the idea that if I took an enemy capital with an elite unit, I was sure to get a leader. It happened in several of my games, but nobody else noticed it. As I kept playing, I realized it was coincidence.

One thing you mentioned has been reported by others. If you have the saved seed feature on and you attack with a vet which promotes to elite, then reload and instead attack with an elite, a leader appears. Now I don't know if that happens or not, but the cause could be that the program generates a number, and the same variable that determines unit promotion determines leader appearance. Since promotion to elite and leader appearance are both rare events... well, I'm sure you get the picture. Again, I don't know the facts on that, but if somebody wants to reload the next time they have a vet get promoted and they have an elite handy, they could clarify things a bit.

 

[Lt. 'Killer' M.]

Ironikinit:
It seems to me that indeed the same number is used, but the threshold for a leader is higher (even with Heroic Epic). So sometimes you get a leader when reloading from a promotion fight, sometimes not.

 

[kring]

It is the same number (if you have preserve random seed on), but as you noted, the probability is not as high so sometimes you will and sometimes you won't.

I produced a scenario with 4 groups of 6 Warriors, surrounded by 6-7 Warrior stacks. I then attacked with my Warriors, using militaristic and nonmilitaristic civs as my civ. I managed to get 2 or more leaders in each instance by doing so. I only attacked the AI Warriors that were on plains/grassland squares. I then immediately disbanded any leader so created since I was testing to see if Warriors could/could not produce Great Leaders. Therefore, my units were not at a significant disadvantage in attacking since the AI units were not fortified or any other bonus, other than the grassland/plains one.

 

[kring]

Here is one of the games attached.

 

[kring]

Hopefully, here is the bic file.

 

[thefrenchzulu]

The threshold is higher and it doesn't always work. I just wondered about it. Would say it works 80% of the time.

I have another theory on leader creation. Without reloading. The question is: When do you attack with the elite unit? I prefer having a stack of elite mounted units. Elites seem to retreat more often.

My theory is based on something like a cycle (wave) It goes up and down. Lets say you attack with a stack and the you have an even contest, e.g horsmen vs spearmen. I find that if the first battle is succesful, taking no hits, it at some stage will go down and the come up again.I find if I attack with the my elite units at the start of an upward cycle (after losing an unit), I seem to generate a leader! Now this is pure theory and I have not tested it to death. Just a theory and it works for me. Anyone want to test it? I also apply to normal veteran battles...

 

[Zachriel]

Originally posted by thefrenchzulu
My theory is based on something like a cycle (wave) It goes up and down.

That is the classic gambler's fallacy. Assuming a random event, each event is independent of preceding events. If you flip five heads in a row, the next toss is still 50-50 (assuming the coin is true).

http://www.nizkor.org/features/fallacies/gamblers-fallacy.html

The human mind will impose patterns where none exist.

 

[thefrenchzulu]

Originally posted by Zachriel

That is the classic gambler's fallacy. Assuming a random event, each event is independent of preceding events. If you flip five heads in a row, the next toss is still 50-50 (assuming the coin is true).

Mentioned that once that I don't have a masters in statistics, but had both calculus and statistics as minors.

If everything in Civ3 was completely random you could not lose a MA to a pike man. Basically no chance of him not taking a hit, but it happens. Regularly. As warmonger I except losses, knowing that it will not continuo. If it did, I would be playing AOE.

I have tested the wave theory a bit using the editor. Once I have had time to test sufficient battles, I will post the results! (At last got time to do testing, but battle with the flipping editor. Not as simple as AOE)

 

[sumthinelse]

Originally posted by thefrenchzulu
I have never tried the new game thing to generated leaders, but have a comment on generating your first leader with an elite warrior. Considering that the jaquar warrior was fortified in a city, gave it various defense bonuses. The probability of winning the battle was thus against you, increasing your change of generating a leader. My unit/battle disparity theory...

If you saw a black cat and then you crashed your car, did the black cat cause the accident? Probably not, but some people remember a black cat (or some other symbolic thing) just before some unusual event happens and remember that occasion. They do not consider that people see black cats thousands of times without anything unusual happening.

Others will objectively study what factors are present at accidents. Intoxication, bad brakes, weather, etc.

I kept track of 138 victories in which elite chariots (attack factor 1) killed spearmen (defense factor 2). There were 8 leaders produced. 128/8=1/16, so this is pretty close to what the expected results would be. Now 138 trials will not give an exact number to determine the real probability of a leader, but I think you need to submit some actual statistics to back up your theory that weaker units attacking stronger units are more likely to produce leaders than the probabilities (1/16 without epic, 1/12 with epic when attacking ) in Mike B.'s post.

 

[kring]

Perception is a big issue. In my games, the AI has been having more wars, with each other as well as me since 1.29f patch came out. In other people's games, it's the opposite.

My point being, that while one person could come up with a string supporting the weak units theory, others could just as easily show it was strong units attacking causing the GL. Would this validate either theory? No. It would just show the limited experiences of the person(s) playing. By that, I meant the experience would be limited to the experiences of the specific players, and not something else. I have played a large number of games, mainly testing mods I have been making, but also to watch the AI behavior in unmodded games. How better to defeat the enemy than to study their tactics? Still, my limited experiences involve just my playing, which can have different results than other players.

 

[sumthinelse]

test