Only 1 leader possible from any given elite

[sumthinelse]

There was a science fiction (or something like that) story that I once read. There was a scientist who heard the conjecture that if you had 100 keyboards and 100 monkeys touching them in random (?) ways they would eventually write all the great books.

So the guy got 100 keyboards and 100 monkeys, let them play for a few minutes, and looked at monkey #1's output. Monkey #1 had written:

"It was the best of times, it was the worst of times...."

(Extra credit for identifying that literary work.)

Improbable phenomena have to occur at some time, so why not now?

Well, after seeing only 27 leaders appear in 418 times with the HE, and writing my last note, on the very next try with the HE I got a leader, and then another one 3 tries later. Today's results have brought the leader creation ratio with HE to 32/471, about 1/15. Still closer to 1/16 than 1/12, but maybe it will even out nearer to 1/12 eventually.

However, the new results without the HE are about the same, 33/361 or about 1/11.

 

[Mike B. FIRAXIS]

Originally posted by sumthinelse

Mike B., is it possible for an army (not the units in it) to be promoted to elite? Never saw one.

Armies are never promoted and do not really have an experience level. (I believe they will be stored internally as either Regualr or Veteran but it is never used for armies).

Mike B., does it look like the odds 1/12 and 1/16 are applied under the correct conditions (before/after Heroic Epic) for generating a leader?

Odds with Heroic Epic = 1/12
Odds without Heroic Epic = 1/16
The odds are halved for defenders (1/24 and 1/32).

 

[Lt. 'Killer' M.]

Originally posted by Mike B. FIRAXIS

Odds with Heroic Epic = 1/12
Odds without Heroic Epic = 1/16
The odds are halved for defenders (1/24 and 1/32).

Ahhh, great to know!
So i guess tit`s implemented correctly - which again makes me feel the RNG is perhaps a bit too streaky for the game.......

Thanx you, Mike!!!!!!

 

[sumthinelse]

Originally posted by Killer


So i guess tit`s implemented correctly - which again makes me feel the RNG is perhaps a bit too streaky for the game.......

I have an idea about this. I figured out how to create a scenario with 25 elite modern armor (with doubled attack/defense stats)attacking 25 regular warriors. In one version it has the epic wonder, in the other it doesn't but the two scenario saves are the same except for the epic.

Without the epic, I got 6 leaders in 91 tries, about 1/15 which is pretty close to the expected 1/16.

With the epic, I got 7 leaders in 75 tries, about 1/11, which is pretty close to the expected 1/12.

If I try this with Killer's game and my game, though, I get strange results. With no epic wonder, Killer and I got 33 leaders out of 361 tries, which is about 1 in 12. So far..

My first theory is that when you attack with a large number of elite units, you get results that are less "wild" than when you attack with a smaller number of elites and keep using the same ones. The large number of units seems to somehow "smooth out" the results so that you get leaders at a rate more like what Mike B. says the odds should be.

Killer, I can upload these 2 scenario saves if you are interested. Each one is only 46K unzipped.

Still, I keep thinking of the scientist who wanted to study a finger. So he cut it off, took it to lab, observed it, and found it couldn't wiggle. In a way, these scenario files are like the finger cut off from its organism. So my second theory is that there is something else in the saved games that the simple scenario does not have, and it affects leader creation somehow.

The third theory is that we just got some strange distribution, like the monkeys that started to write "A Tale of Two Cities," and eventually even Killer's and my games will start behaving as expected.

 

[Lt. 'Killer' M.]

@ sumthinelse: gimme the savs!!! I have extremely little time at the moment, but this one is burning under my nails so to speak.

I do think the games will even out, I remember a time when I couldn`t play a normal game because I got too many leaders, like one every few turns even if only defending. I once had the Pentagon build before Chivalry because I ran out of Wonders to use the leaders on, so I build armies and then blew a leader on a small wonder. If I add this to the current games - well....

 

[sumthinelse]

OK Killer, the zip file here has 2 scanario saves:

noepic.sav is without the wonder

hasepic.sav has the wonder.

In both you attack 25 reg warriors with 25 elite modern armor. There is not much else on the map. I would suggest you attack one at a time because if you move the whole stack at once you can't tell when the leader got created.

My latest results are posted above, and they agree with Mike B.'s statement about the odds. So far.

 

[sumthinelse]

Originally posted by Mike B. FIRAXIS


Armies are never promoted and do not really have an experience level. (I believe they will be stored internally as either Regualr or Veteran but it is never used for armies).



Odds with Heroic Epic = 1/12
Odds without Heroic Epic = 1/16
The odds are halved for defenders (1/24 and 1/32).

Mike, thank you for you time and your very reliable answers.

 

[Loopy]

Originally posted by Mike B. FIRAXIS
The odds are halved for defenders (1/24 and 1/32).

I love how tidbits like this trickle out. I don't think I've seen this mentioned elsewhere. Or is it?

 

[sumthinelse]

Originally posted by Loopy


I love how tidbits like this trickle out. I don't think I've seen this mentioned elsewhere. Or is it?

First time I have seen it. Mike is just trying to help, though. Let's not shoot the messenger.

 

[MuddyOne]

Sumthinelse,
The "smoothing out" you are noticing is what happens when you have a larger sample set. Flip a coin 10 times and you are more likely to have a 6/4 outcome than a 5/5 outcome. Flipped 100 times and it moves closer to 50/50, a thousand even closer.

 

[sumthinelse]

Originally posted by MuddyOne
Sumthinelse,
The "smoothing out" you are noticing is what happens when you have a larger sample set. Flip a coin 10 times and you are more likely to have a 6/4 outcome than a 5/5 outcome. Flipped 100 times and it moves closer to 50/50, a thousand even closer.

I know exactly what you mean, but that is not what I was talking about. I got a "smoother" result from a smaller number of trials. What I was saying is that I got "rough" results from a sample set of 361 total tries using a few units over and over again, but when I used a larger number of units I got a "smoother" result from 75 total tries. Could be an anomaly. Or not.

This would be like getting "rough" results from flipping a single coin 100 times (37 heads and 63 tails) but getting results like 9 heads and 11 tails by flipping 10 coins twice. If the single coin was irregular, that might make sense.

There is a story (probably made up) about John D. Rockefeller asking a porter on a train how much was his "average" tip. He replied that his "average" tip was $2.00, a very big tip in those days. Rocky gave him $2.00 and commented that if that was his average tip, he must be doing pretty well. The porter replied, "Not really. That's the first 'average' tip I've had in 2 months."

"Random" numbers behave kind of strangely sometimes.

 

[sumthinelse]

Killer, I found the flag in the unit structure that indicates whether an "elite" has been used! First I have to decompress the file, then find the unit structure with a hex editor.

I took one of the scenario files I posted here, hasepic.sav, which was producing leaders at about 1/12. I set the "used" flag with a hex editor in each elite modern armor unit and saved it as a different file, and from this file I can't create a leader in 100 tries. So I am confident I have found the "used" flag.

Don't worry, the files I uploaded here do produce leaders. I didn't alter them.

This is not for cheating, although I could easily turn the flag off in any unit. It's to figure out which elite units are "used" so I will know which ones to attack with to avoid "leader droughts" and know when to build new veteran units etc. It's really tedious for me to try to keep track of which elites are used. This tecnique should help.

 

[Lt. 'Killer' M.]

WOW, sunthinelse, this is HOT NEWS!!!!!!!

I'd never know how to do somehting like that, and if someone showed me (I did stuff like that in my Amiga days ) I'd probably scream from frustration after 5 min....

You have earned yourself 3 thumbs up:

 

[Loopy]

Originally posted by sumthinelse
First time I have seen it. Mike is just trying to help, though. Let's not shoot the messenger.

Hrm, I guess you could read what I wrote as sarcastic though I wasn't thinking that way when as I was typing. Let me rephrase that -- I'm really glad we've got a Firaxian programmer shedding light on gameplay nuances.

 

[sumthinelse]

About finding the bit in the structure:

Originally posted by Lt. 'Killer' M.


I'd never know how to do somehting like that, and if someone showed me (I did stuff like that in my Amiga days ) I'd probably scream from frustration after 5 min....

Yeah, and even now that we know which one it is, this is not going to help everybody, unless I can write a program to change the color of these units or something like that.

But if you get confused about which ones are "used" in one of your games you can send it to me and I will analyze it.

But another problem is that when you have a big stack of units, I don't know which one would appear in which order in the game graphics. In the un-compressed save file, each unit has an ID number. Maybe I could figure out how to display that. Gramphos displays the ID number in his CIV3 Multitool saved game editor and units appear in their order in the "Loaded On" field in the "Units" tab.

And then there is the problem that CIV3 does not give us the option of displaying the coordinates on the map squares, so I would have to tell you "down 25 and left 38 from Scheissburg" to tell you where a unit was. Since they have the map rotated 45 degrees I would have to explain what "down" and "left" mean. I guess you would need the CIV3 Multitool also unless I can find a better way.... Sigh.

 

[Lt. 'Killer' M.]

sunthinelse, I guess this is just too much work to undertake! If it goes as I suspect it goes, the next patch will change this if you write something to help everybody, but if you save yourself the hassle, they`ll keep things as they are

I know, I`m a hopeless pessimist

but thanx a lot again for finding this! At least, we know now that something can be done about it...

Oh, before you write a program to chenge the color of the unit - wouldn`t it be easier to write somthing that simply resets the flag?????

 

[sumthinelse]

Originally posted by Lt. 'Killer' M.
Oh, before you write a program to chenge the color of the unit - wouldn`t it be easier to write somthing that simply resets the flag?????

It's pretty easy to turn the flag off with a hex editor, but then I would be "cheating." In fact, I guess it would be too easy to win that way, the opinions of others notwithstanding. I can tell you how to do it though. For research only, of course

I just hacked a save file and CIV3 crashed when it tried to load it CIV3 blew up That must mean I'm making progress.

Killer, is there a way to attack with an entire stack by using a "go to" command? I tried "j" and it didn't work for attacks....

 

[sumthinelse]

Originally posted by Loopy

Hrm, I guess you could read what I wrote as sarcastic though I wasn't thinking that way when as I was typing. Let me rephrase that -- I'm really glad we've got a Firaxian programmer shedding light on gameplay nuances.

OK, I just don't want Mike to get the wrong idea. Sounds like he is sympathetic, but of course he can't give us everything we want. And if he did the game would be too expensive, too big, too slow, and, #1 on the list:

Too hard to debug!!!!!

 

[damunzy]

sumthinelse
Hopefully Firaxis will make the ELITE units visually different someday. Is there a way to delete the unit from the save so you don't end up accidently reusing that ELITE unit and wondering why you never get any leaders? That wouldn't be cheating b/c you would be just disbanding the troops.....

I've suggested this to someone else for his findings and thought I would suggest it to you guys also:

I suggest that you post your findings in Tutorials, Reference, & Guides.

 

[IDSmoker]

Since most of the current GL related questions seem to be answered (either by Firaxis, or by posters with way too much time on their hands... way to go Killer! <grin>), I thought I would throw another one out there, namely....

Can units with the 'blitz' attribute (the ability to attack more than once per turn) create GLs on any attack other than their first?

My anecdotal evidence as the Germans (I've been having a LOT of fun with Panzers! <grin>), is that they can not, but I haven't done the kind of structured research needed to prove it. I'm generally able to create a GL every other turn or so (during almost constant war). But I have not seen a single unit create a GL on anything except it's first attack.

A related question would be:

Can a blitz unit that gets promoted on it's first attack, create a GL on it's subsequent attacks?

Again, this is something that I've never seen happen, though it would be rather more unlikely.


On other note...

Originally posted by sumthinelse

It's pretty easy to turn the flag off with a hex editor, but then I would be "cheating." In fact, I guess it would be too easy to win that way, the opinions of others notwithstanding. I can tell you how to do it though. For research only, of course

sumthingelse: Great going!!! Depending on your ability as a programmer (and if you aren't a programmer, you need to start thinking of changing careers! <grin>), you should think of creating a little utility that will flip the appropiate bit, assuming it won't take much effort. Then if Firaxis doesn't fix the display issue of these 'used' elite units, you can consider creating a more complex utility that will flag/report these units for the user.

I know it's a cheat, but resetting the 'used elite' flag would be a quick and easy solution to the current problem. Until multiplayer gets implemented with the expansion pack (which will probably also change the ".SAV" file, invalidating the quick & dirty utility), the only person harmed by cheating is the cheater.

My personal philosophy is that I would only use this type of utility in those instances when I am unable to determine which unit created the GL. I currently upgrade, disband, or put the 'Used Elite' unit in an army depending on circumstances. If I can't figure out which unit created the GL (others have well documented what circumstances can create this situation), I put all the 'tainted' units through the grinder.

It would be really nice to have another option in these types of situations! <grin>

Whatever you decide to do with this hard-won information, I hope that you will at least post the hex-offset value, so that others can make use of it.

Thanks for the hard work!