Got my first leader, some advice please..

[vmxa1]

All of this seems to verify that holding a leader for a long time makes no sense, unless you are not going to be at war. Even then I say it is like having a large sum of money and not putting it in the bank. Late in the game I use leaders to rush temples in new towns, just to get them out of the way. I am sending elite armour to city after city so I get leaders to burn. When I use to hold them, I did not get more. I do not need them for wonders now as I have them all and the few left can be make in a short time in a city that is now doing nothing anyway (or I may rush a leader back).

[Harovan]

Originally posted by Dimension
I suppose I'll start keeping track next game, though

So will I, if I ever will have enough time left to play another one

[Dimension]

Yeah, those should probably be else if's, because that would give a 0.13% chance (1 in 768) of generating 2 leaders if you had none, and a 0.0027% chance (1 in 36863) of generating 3.

Anyway, it would sure take a lot of data to make any sort of conclusion (like tell if it was 10.4% or 11.1%). Those make the most sense to me, but Firaxis could be putting their numbers together in some completely different way.

I suppose I'll start keeping track next game, though

[Harovan]

Originally posted by Dimension

if ((!Leaders) && (!(double(rand())/RAND_MAX)*15)) {
      MakeLeader();
   }
if (CivTrait(Militaristic) && (!(double(rand())/RAND_MAX)*47)) {
      MakeLeader();
   }
if (GotWonder(HeroicEpic) && (!(double(rand())/RAND_MAX)*47)) {
      MakeLeader();
   }

Did you make a reverse engineering, huh? Don't you know that you are not allowed to?? Nevertheless your code is wrong, because it would give some odds to create three leaders out of the same battle!

But I must agree, that having 2 leaders at once, while I think it's possible (so many have reported it and they can't be all liars), is neither predictable nor are the odds to create the 2nd as high as for creating the 1st. I think Firaxis will not make a statement to prove our guesses.

We need a lot of numbers (fights with elite units involved, with or without existing leaders, etc.) to calculate the odds ourselves. It would be great if the results could be reported and collected in one thread (and be it this one), so that everyone could make his theories out of the same raw data base.

[Hurricane]

Sorry. I got your point, but I just wanted to continue the discussion by provoking people to do some more testing. You know how easily a number such as that is taken as a fact.

[Dimension]

Man, talk about missing the point entirely.

I said it was possible, but I'm absolutely sure that's not what the actual code looks like. It's just a scenario that leaves room for a small chance of a 2nd leader. You could just as easily modify it so that there is a 2% chance of a 2nd leader. My point is that, while Sir Ralph's theory about multiplying by 4/3 makes sense, it doesn't leave as much room for having 2 leaders at once.

And I also think this whole 2-leader thing is pretty suspicious...

[Hurricane]

I decided to test a bit to finally with my own eyes see two leaders at the same time. So I loaded a savegame where I recently had got a leader and started making attacks with elite units (with the leader sitting idle at my capital).

After 104 succesful elite attacks (including 4 succesful defence combats) I got a total of 0 leaders.

Before doing this test, during regular gameplay, I had immediately used up the leader, and then got 3 new leaders (which I all used up as soon as possible).

So I´d say that the chance of getting two leaders at once is much lower than the 4.167% Dimension suggests.

And I still haven´t seen a savegame or even a screenshot with two leaders.

[Dimension]

...and actually, the fact that you have a small chance of generating a 2nd leader while you already have one makes me lean towards it being additive, because you could get a scenario like:

if ((!Leaders) && (!(double(rand())/RAND_MAX)*15)) {
      MakeLeader();
   }
if (CivTrait(Militaristic) && (!(double(rand())/RAND_MAX)*47)) {
      MakeLeader();
   }
if (GotWonder(HeroicEpic) && (!(double(rand())/RAND_MAX)*47)) {
      MakeLeader();
   }

...which would give a Militaristic civ with the Heroic Epic a 10.417% chance of generatic leaders normally, but a 4.167% chance of generating leaders after they already had one.

[Dimension]

Well, if Firaxis has verified the numbers 1/16 and 1/12, then either way could be correct, whether it's additive:

base leaders:  1/16                        (6.250%)
militaristic:  1/16 + 1/48 = 1/12          (8.333%)
heroic epic:   1/16 + 1/48 = 1/12          (8.333%)
mil. + heroic: 1/16 + 1/48 + 1/48 = 5/48  (10.417%)

or multiplicative:

base leaders:  1/16                        (6.250%)
militaristic:  1/16 * 4/3 = 1/12           (8.333%)
heroic epic:   1/16 * 4/3 = 1/12           (8.333%)
mil. + heroic: 1/16 * 4/3 * 4/3 = 1/9     (11.111%)

1/9 would be easier to remember, and it'd be nice to have that extra 1%, but I'd still like some real verification of how they stack.

[Harovan]

Originally posted by Dimension
Are the resulting ratios additive (1/12 + 1/12 = 1/6 total, or 16.67%)?

I disagree with that and with your numbers below.

1/12 is 4/3 of 1/16. So the result of having Heroic epic and being militaristic should be 4/3 of 1/12 = 1/9 or 11.11...%.

[vmxa1]

Those are the numbers from Firaxis.

[Dimension]

3 in a turn? Nice.

I'd still like to see some solid numbers on leader generation. I've heard several people say chances are 1/16 or 6.25% of getting a leader (from elite units in non-barbarian combat, of course). I've also heard several times that your chances are 1/12 or 8.33% after you build The Heroic Epic. People also say your chances start off at 1/12 if you are Militaristic.

Has anybody in any way verified these numbers with Firaxis, with game code, or with extensive testing? How does the Military / Heroic Epic bonus stack? When you have both, are they they redundant (1/12 total, or 8.33%)? Are the resulting ratios additive (1/12 + 1/12 = 1/6 total, or 16.67%)?

Having played nothing but the Aztecs lately, I would have to say that I doubt it is 16.67% for both. What makes the most sense to me is that the bonuses themselves would be added. This could be done as follows:

base leader production = 6.25% (1/16)
militaristic = 6.25% + 2.0833% = 8.33% (1/12)
heroic epic = 6.25% + 2.0833% = 8.33% (1/12)
mil. + epic = 6.25% + 2.0833% + 2.0833% = 10.4167% (5/48)

This could be better expressed as:

base leaders:  3/48                       (1/16 or  6.2500%)
militaristic:  3/48 + 1/48 =  4/48        (1/12 or  8.3333%)
heroic epic:   3/48 + 1/48 =  4/48        (1/12 or  8.3333%)
mil. + heroic: 3/48 + 1/48 + 1/48 =  5/48 (5/48 or 10.4167%)

This seems logical, although I would bet it's still not how it works, or at least the numbers are off a little bit, because from a code standpoint, 12ths (and therefore 48ths) can't be expressed normally with byte, word, or double-word ratios. That is, 12 is not a factor of 256, 65536 or 4294967296. Firaxis could've coded these bonuses as type real, with a value of 0.0069444, but that seems odd.

The way C++ works, you predefine your max random integer, then use that for all random numbers generated. Generally, you're going to define RAND_MAX (+1) as either a power of 256 or a power of 10 (i.e. 65536, 100000, or something like that). That's why it seems strange that 48 is not divisible by any power of 256 or any power of 10. I doubt all random numbers in Civ3 are based off of factors of 48. It is common to use the following kind of statement to find a random number with max value n:

(double(rand())/RAND_MAX)*n

That works just fine, and may be the case with Civ3, but it just seems strange. Software that makes extensive use of random numbers will generally not use such an arbitraty inefficient method to generate those numbers.

Perhaps they are coded like this:

base leaders:  16/256                             (6.2500%)
militaristic:  16/256 + 5/256 =  21/256           (8.2031%)
heroic epic:   16/256 + 5/256 =  21/256           (8.2031%)
mil. + heroic: 16/256 + 5/256 + 5/256 =  26/256  (10.1563%)

That still seems a little odd, though. Anybody know how this actually works?

What I'm trying to say here is that I wouldn't be surprised if the whole theory of militaristic/heroic epic giving you a 1/12 chance of leaders was completely wrong.

[Aeson]

Regardless of whether or not 2+ active leaders are possible, having an active leader lowers the chances that another will be produced. I was able to get 3 leaders in one turn once. I had used each leader as soon as it was available though, before the next battle was fought. I reloaded to see if using the leaders had made a difference. When I didn't use the first leader, I didn't get any more. This was using the exact same series of battles that produced the 3 leaders the first time.

[Dimension]

Interesting... So, does the Military Advisor screen switch from "available leader" to "available leaders"?

This might change my strategy quite a bit. I generally time my wars around getting tech for key wonders, and don't build my first army unless there are no wonders available. If I could hold onto two or three leaders at a time, I could plan things a little better.

I always use my leaders ASAP, though, and in my current game, I just got my 7th leader in 780AD ... so I can't help thinking it's better to use them immediately.

[Harovan]

So many people reported to have had 2 leaders at the same time, but nobody thought of a screenshot or a savegame. Too bad. I didn't have so much luck yet, and as a builder player probably never will, but if I ever will see 2 leaders in a row, I will save the game (A screenshot is no real prove either).