a few observations on leader generation

[Slacker]

I've been playing as the Germans and Russians and not getting any leaders either. Granted, I've not been overly aggressive, but I've fought a decent number of wars. It also seems the AI doesn't get them or use them either. I've never run across an AI army before.

This might also sound like a stupid question, but can you get a leader fighting barbs? It seems like it wouldn't, but it'd be nice to have confirmation.
Nice holiday themed smilies, btw

[Hurry]

Then the chance of getting a second one at the same time is considerably lower. For example: getting leader one 1/16, and getting another while still having the first leader unused 1/32.

[karlmarx9001]

'I am almost CERTAIN you can have only one unused leader at a time. I'd strongly recommend using them ASAP, especially if the war isn't over yet. First leader should, I think, ALWAYS be converted into an army. If you're fast, you can do that in one turn, and keep attacking to get another!'

Huh? I've had two active at once, not turned into an army or anything, just sitting in my capital, so there's no "cap" on the number of leaders in your possession.

[Hurry]

you should run a test with caching-in the leader (converting it to an army, or otherwise getting rid of it) to see if the problem is having more than one un-used leader at the same time.

Ok, thanks for the tip . I did the test. It turned out just as you suspected. If i immediately used my leader from event 16 to rush a building, I would get another one on event 29.

So, I would say this is the fact until someone come up with proof of having several leaders at the same time:

YOU CAN HAVE NO MORE THAN ONE LEADER AT A TIME.

The savegame is attached. Do these moves to test for yourself.

1: attack the cavalry with your elite. You get Leader 1.
2: attack Ur with a veteran armor.
3: attack Ur with a regular armor.
4: attack Ur with an elite armor. You´ll get a leader if you used the first one up, otherwise not (too bad for you . )

[Thrawn05]

I've been playing the game for a couple of weeks none stop, and I tend to always play as the Zulu, but not once have I ever made a leader. I have destroyed many civs trying to cough one up, but I just can't. All I get is a squad of elite warriors and Impis.

[Nakar Gabab]

I am almost CERTAIN you can have only one unused leader at a time. I'd strongly recommend using them ASAP, especially if the war isn't over yet. First leader should, I think, ALWAYS be converted into an army. If you're fast, you can do that in one turn, and keep attacking to get another!

[shuttleswo]

Originally posted by Hurry
In event 29 I would ALWAYS get a Leader if I attacked with an elite unit. A non-elite unit would always be upgraded. However, if I already had got a leader in event 16, a new Leader would not appear in event 29. If the only thing considered was if the random number falls under the 1/16 chance, I would get two leaders during my Test2 run. I did not. Only one leader.

you should run a test with caching-in the leader (converting it to an army, or otherwise getting rid of it) to see if the problem is having more than one un-used leader at the same time. i KNOW i have gotten more than one leader in a single turn, but i rushed the first leader back to a city and hurried a wonder before getting the second. if i had the save game for it i would test myself.

i'm inclined to think that only one leader is allowed at a time, the availble LEADER window on the mil advisor screen also seems to imply this.

[MadWombat]

... or the probability is not 1/16.

[Hurry]

My point is till valid: there must be more than a fixed probability of getting a leader. Look at my test again.

In event 29 I would ALWAYS get a Leader if I attacked with an elite unit. A non-elite unit would always be upgraded. However, if I already had got a leader in event 16, a new Leader would not appear in event 29. If the only thing considered was if the random number falls under the 1/16 chance, I would get two leaders during my Test2 run. I did not. Only one leader.

So either people are wrong in that you can get several leaders shortly after each other (same round), or then the probability is not fixed.

[karlmarx9001]

If I remember correctly, I think someone from Firaxis did. They said there was a 1 in 16 chance of a leader appearing anytime an elite land unit wins a battle, with a 1 in 12 chance for militaristic civs.

Or maybe it was from the strategy guide. Can anyone confirm this?

[Nadexander]

I wish firaxis would just tell us what the equation for deciding the chance a great leader would be spawned instead leaving us quessing! Just a line of code... just one... not enough to reverse engineer anything .. i promise!

[karlmarx9001]

Well, here's an example of why leaders are so hard to get:

Most of my combats are either knights vs. pikeman, cavalry vs. musketmen, or tanks vs. rifleman, with the defenders in size 7-12 cities. According to the combat calculator:

http://www.columbia.edu/~sdc2002/civulator.html

The knights, cavalry, and tanks have a 53%, 60%, and 86% chance of winning, ignoring the fact that some combats will be fought against defenders with less than the full number of hit points for the moment. Dividing by 16, 3.5%, 4%, and 5.5% of elite units should produce units, according to the era of the combat.

That's 1 in 30, 1 in 25, and 1 in 20 of all elite combats. Sounds fine to me.

[Nakar Gabab]

I took about 30-40 cities in a single turn the last game I played; I got one leader out of it, and many promotions from veteran to elite. I was using armor, so many of the elite units were attacking after being promoted... but I still didn't notice anything major.

Is everyone SURE you can get more than one leader at a time? I was the Aztecs and I'd figure that sacking dozens upon dozens of units in a turn, with Heroic Epic and the Militaristic trait, would produce 2-3 leaders if it's possible to have that many at once...

[karlmarx9001]

'As you can see, this is a test on a very small scale, and I can only say that it is less probable to get a second leader if you recently got one. As some have pointed out, several leaders can be spawned the same turn, so there must be something more complex involved than just a simple "No More Leaders In The Next 10 Turns For You"-routine. It might be that there is a sliding percentage. I had not got a leader for a long time before this test, so it might be that the chance of spawning a leader was the same as for upgrading. When I got one, it could have jumped up again to a much lesser probability.'

As the chance of a getting a leader for winning with an elite is only 1/16th, what you're seeing is probably random chance. The chance of generating a leader is so low that you'll see inconsistent results over a small sample set.

Here's my speculation as to how its all generated:

The game uses a psuedo-random number generator which produces a random number, given a seed. The seed is saved in the same game. Every time a random number is generated, the seed is slightly changed based on the action being taken, then applied to the function, and a percentage result is produced (50%, 73.4%, whatever). That number is then used for whatever action just occured. This explains why you get the same results on save/reload, but if you do actions in a different order you get different results. That the seed is modified by the details of what you're doing (map square, probably, among other things) keeps you from reloading and attacking under the same conditions elsewhere in the game, and getting the same result, which probably looks a little too deterministic for players.

The implications of this system are that if you're willing to waste the time, you can reload a save game over and over and do things in a different order until you get an "optimum" result. Also, diplomacy appears to involve some random generation, so that can also be used to advance the seed. It just stops save/reloading every time you lose a battle.

To really test leader generation, you'd probably need to have something like 128 elite units in a row win, not just fight. You can't use "promoting veterans" as part of it, as the chance of promotion from veteran to elite seems to be much higher than 1/16th. Now that I think about it, this wouldn't be that hard to do, although boring.

[Hurry]

In my test I was waging a war against the Babylonians (I am the Greeks). I have Armor, Artillery and Infantry, while the Babs have Cavalry and Infantry and just got Artillery. I am now attacking several of his cities and have little trouble. Anyway, this is how I tested.

I recorded the exact outcome of each battle, and also recorded the number of random events needed for each battle. For example, if my veteran armor attacks a conscript infantry in mountains, the battle outcome might be Hit (on the infantry), Wound (on my tank), Hit (and the Inf dies). This requires three random numbers. Another one is needed to decide whether the tank is upgraded or not, making a total of four events.

After recording a number of events (about 40), I reloaded and made my attacks in a different order with different units and again recorded the results. As you know, the random seed won´t change by reloading, so results can to some extent be predicted.

EVENT__UNIT_HPs_DEFENDER ___HPs_OUTCOME_____EXTRAS
1-9_____Arm__5__Inf, fort, city ___3__NWWWHHWH__Not upg
10-12___Arm__4__Inf, fort, city ___2__HH___________Not upg
13-16___Arm__5__Inf, city_______2__WHH __________UPG/Leader!

TEST2:
1-4_____Arm__5__Inf, fort, city ___3___AWWW______Retreated
5-7_____Arm__5__Inf, Mountn. ___2___HH__________Not upg
8-14____Arm__4__Inf, fort, city ___3___CWHHWW____Def not upg
15-16___Arm__5__Cavalry,clear ___1___H___________UPG/Leader!
17-22___Arm__4__Inf, fort, city ___2____HWWWH_____Not upg
23-26___Arm__3__Inf, city _______2____HWWW_______Def not upg
28-29___Arm__5__Inf, fort, city____1____H____________Leader*

Event: Number of random events
Arm: Armor
HPs: Number of hit points
Defender: Type of defender (Inf=Infantry, Cav=Cavalry) and terrain (fort=fortified, city=metropolis-sized city)
Outcome: N=not hit by cannon, W=attacked wounded, H=defender wounded, A=1 hit by defending artillery barrage, C=1 hit by defending cannon barrage.
Extras: Retreated=faster unit retreat from battle. Neither side have the chance to upgrade, Not upg/Def not upg=unit not upgraded (def when defender won battle). Elite units which would have been upgraded instead spawn a Leader.
* See below

The interesting thing concerns event 29. If I had got a Leader from event 16, nothing would happen if I used an elite unit at event 29. If I here had a veteran or less experienced unit, it would get upgraded. But, if I used a veteran or less experienced unit at event 16 (which, thus, would be upgraded and not spawn a leader), and an elite at event 29, a Leader would be spawned from this battle!

As you can see, it is the event number (or actually, the random number residing at that event) which decides when a unit will be upgraded - not anything like hard or thight battles, or capturing the enemy capital. In Test2 my elite Armor spawned a leader after killing a 1HP Cavalry in clear. Not exactly a hard-fought win!

As you can see, this is a test on a very small scale, and I can only say that it is less probable to get a second leader if you recently got one. As some have pointed out, several leaders can be spawned the same turn, so there must be something more complex involved than just a simple "No More Leaders In The Next 10 Turns For You"-routine. It might be that there is a sliding percentage. I had not got a leader for a long time before this test, so it might be that the chance of spawning a leader was the same as for upgrading. When I got one, it could have jumped up again to a much lesser probability.

Sorry for the long post, but I felt this needed some thorough explanation.