The mystery of Great Leaders, and the randomizer

[gnome]

I have heard it said that there is a 1/16 chance of an elite unit generating a Great Leader in battle if attacking and winning, and a 1/32 if defending and winning.

I have found this to be far from the truth in most of the games that I have been playing... my most recent game, I engaged in a major war early in the game... had dozens of battles, none of which generated a leader. There have been several massive wars since (that fortunately I have not participated in)... and I have seen exactly one army in the entire session, used by the computer.

Most other games, there have been no great leaders at all.

There is a lot of speculation about how to improve the chances, but a lot of them seem to resemble people that have little rules about how to pull the arm on a slot machine--nothing seems to work consistently for everyone. I wonder if we could get to the bottom of this issue... there is something I would like to note... when I installed Civ3 on another machine, I got great leaders in almost every game, with the same playing style. Unfortunately I only played a handful of games on this machine as it is unstable.

Is it possible that there is something specific about a computer's setup that determines whether great leaders can happen? This would also account for some posters claiming to have no trouble creating leaders, but others finding it almost absent from their Civ experience.

The other possibilities are, it seems to me:

1. The 1/16 figure is just incorrect--perhaps someone from Firaxis can speak on this?-- and those who commonly get great leaders are just lucky or fight a LOT more battles, somehow.

2. There is some undocumented factor at work here, some other requisite that those players who have trouble forming great leaders have innocently neglected.

3. Something wrong with the randomizer. I have often suspected this, particularly in battles that do not seem particularly random. When fighting a low-tech unit with a high-tech unit, I do not mind if it kills me occasionally...but it should often get just one or two hits... and this seems rare... Sometimes it seems like hit points are almost irrelevant--if a unit gets hit at all, it will die or retreat at 1 HP. Maybe not always, but more often than can be accounted for by random chance. Does it seem to anyone else that there is some nonrandom element going on? Maybe some statistical studies from observing units can resolve this--or maybe we can hear from people that have upped the hit point levels of their units in a mod.

I propose that some of the people (and I will do this myself) that have had trouble creating great leaders, try installing it on a different machine, and see if their experience is different. That way we can rule out playing style, and maybe have something to show to Firaxis as a problem.

[Cyclotron]

I think it's unlikely that your computer is the reason for what you consider skewed results. I would think that simple luck as well as your way of viewing the results are to blame for your apparent troubles. I seem to get 3 or 4 leaders a game when I carry through to the modern era.

[gnome]

But my question is, how can simple luck account for it, even a 1/16 chance should come up once or twice in several dozen trials...

[Cyclotron]

Well...

1) How many of those battles involved elite units?
2) Are you using v. 1.29? If not, you may be using some elite units a lot that have already produced leaders, and not know it.
3) Are you a militaristic civ in these games?
4) And just to make sure... you have no active leaders when you are fighting, do you?

By the way, I think the 1/16 applies to either militaristic civs or the heroic epic... I thought that the base chance was 1/32. But I could be wrong.

[Theseus]

Base chance is 1/16 on attack, improved to 1/12 with Heroic Epic. Militaristic only increases the chances of promotion, and has no impact whatsoever on GL generation.

gnome, this has been discussed a LOT... your results are waaaay inside of the statistical safe zone. Poke around here and in the Strategy forum.

I'm pretty sure that there aren't any more hidden factors.

[Edit: Corrected typo: 1/6 -> 1/16]

[gnome]

Originally posted by cyclotron7
Well...

1) How many of those battles involved elite units?

Lots of them. They were created by cities with Barracks, and reached Elite status after the first couple of battles they fought. I created vast numbers of them, and each fought several battles.

2) Are you using v. 1.29? If not, you may be using some elite units a lot that have already produced leaders, and not know it.

I am using 1.29.

3) Are you a militaristic civ in these games?

...
By the way, I think the 1/16 applies to either militaristic civs or the heroic epic... I thought that the base chance was 1/32. But I could be wrong.

No, but my roommate (on the same machine) has tried militaristic civs with the same results. And even if the chance is 1/32, I still feel like I am not even getting odds that good.

4) And just to make sure... you have no active leaders when you are fighting, do you?

Correct... I am failing to even get a first leader.

Originally posted by Theseus
Base chance is 1/6 on attack, improved to 1/12 with Heroic Epic. Militaristic only increases the chances of promotion, and has no impact whatsoever on GL generation.

gnome, this has been discussed a LOT... your results are waaaay inside of the statistical safe zone. Poke around here and in the Strategy forum.

I'm pretty sure that there aren't any more hidden factors.

This is my contention, that I'm way OUTSIDE the statistical safe zone. I'm not sure what you mean by 1/6 and 1/12... if the probability went to 1 in 12 from 1 in 6, then building the Heroic Epic should be the LAST thing you would want to do--did you reverse those?

And I have read lots of the other threads about this... none of them seem to have consistent information in them...

[Theseus]

Sorry, I corrected the typo... it's 1/16 without the HE.

I think we've gotten a consistent view on the mechanism... give me a little bit and I'll try to find a relatively recent thread with all known factors.

[gnome]

Originally posted by Theseus
Sorry, I corrected the typo... it's 1/16 without the HE.

I think we've gotten a consistent view on the mechanism... give me a little bit and I'll try to find a relatively recent thread with all known factors.

Thanks, I'll check it out when you find it.

[Theseus]

Sorry, I just took a quick look for it, no joy.

Here's what I can remember of the top of my head:

1/16 chance on attack by an elite, improves to 1/12 with HE.

1/32 chance on defense by elite, imrpoves with HE to 1/24 (I think).

Armies cannot generate GLs, but if an elite unite defeats an Army (on attack or defense), it can do so, at normal odds.

No GLs from air or naval combat.

Promotions: Any unit that wins two combats in the same turn is automatically promoted. This is relevant in that it *seems* that GL generation is a subset of the promotion function, and if so, then MA, with 3 attacks in a turn, *might* be able to generate GLs more often.

Once an elite generates a GL, it is named, and until upgraded, denoted with a *. If it is upgradeable, then when upgraded it is demoted to veteran and the * is removed, and can thus be promoted to elite again, and generate another GL. No info on whether these upgraded units have any greater chance of generating another GL.

That's it. None of the stuff about modernity of units, first to fight, last hp, in / out of a city, etc., matters.

MY advice: Fight a LOT. Conserve your elites for easy pickings, but use them as much as possible. Consider sending a GL-generation team into the field... 1 or more of: strong elite attackers (preferably fastmovers), strong defenders, and a couple of bombard units. Put'em on a hill or mountain, and keep banging away until you get some GLs.

[Blake]

I used to be one of those people that never got GL's (and I didn't change machines!), but that was from not using elites enough . Anyway, in a recent game in the first 20 or so elite battles I got no GL... was getting pretty disheartened. At this stage I was still thinking there was extra factors like a battle which captures a city having a better chance of generating a GL...

Then I attack a spearman escorting a settler out in the open, and get a GL, which I i used to rush FP. Next unit activated was an elite archer which I used to attack a swordmen in the open and generated another GL.

Around 40 elite battles later I got my third GL, from an elite rider.

It's all just luck, pure and simple. Sometimes you have bad luck, sometimes you have good luck. Sometimes you have extremely bad luck, sometimes you have extremely good luck...

If a thousand civ players each fight a hundred elite battles, statistically you can expect a few to not get any leaders at all from those hundred battles.

gnome, I am afraid that you are simply one of those few.
(fortunately that doesn't alter your chances of the next battle generating a GL...)

[Flight]

in the game i just completed i had precisely 1 leader which i used to rush teh FP
this after several civ-killing wars taking over half the landmass

in the game previous to this i fought in exactly the same way and conquered roughly the same amount of land with roughly the same number of combats and therefore chances of generating a leader but i got something like 9 leaders throughout.

It sure is pretty weird, and from just this i'd say not dependant on some computer-specific variable

[Demerzel]

i got 8 GLs in my last game but I spent most of it waging war against my emperor level AI opponents. I tend to emphasis the usage of Elite troops when attacking & their subsequent protection when they are injured, in order to maximise my chances.

so that's 15 GLs in the last two games, both won on emperor level via domination as space race is a girlie's way of winning

[Hurricane]

Some guys over at the other forum did some extensive tests on leader generation, and after it was discovered that an elite only can generate one leader, the results were exactly as predicted. That is, that the chance of getting a leader is indeed 1/16 (or 1/12 with HE).

[Grumbold]

I've had very inconsistent results with the leaders too, but all within acceptable probabilities. All my early games were playing as the Romans and yet it was several games before I saw my first GL. After that the trend swung the other way for a while and I had several games with more leaders than I could find uses for playing a non-military civ. Go figure.

[Zachriel]

The odds

When an event is relatively rare (1/16), and the number of trials is low, then results will vary considerably. Here are some examples (assuming non-militaristic civ):

With 16 trials
none 36%
exactly one 38%
at least one 64%
at least two 26%
at least three 7.4%
more than three 1.5%

With 32 trials
none 13%
exactly one 27%
at least one 87%
at least two 60%
at least three 32%
more than three 14%

Considering just the "none" category: If twenty players win 32 successful elite combats each, then it would not be unusual if a couple of the players will have generated no Leaders.