leader drought

[chiefpaco]

Originally posted by sumthinelse
There may be other situations in which an elite victory cannot produce a leader. My elite units have killed 200 Zulu Impi (when I had no leader) without creating a single leader! Odds: about 1 in 400000! When I attack Zulu archers and warriors I get leaders except in the cases stated above.

Has anybody ever generated a leader by killing a Zulu Impi?

My friends, help me to observe what conditions allow or prevent leaders.

I am just making sure you actually killed the Impi. If they retreated & did not "die", then it is not counted as a victory. It seems like this is the case, & you know that, I'm just making sure.

From what I've experienced, I don't think there is any other factor, other than the chance of a leader occuring %. I've generated leaders under all circumstances, & none stand out to me, at least.

Long droughts can occur. There is no way for us to prove any certain circumstances change the likelihood though. I've had some games where I generated more leaders than I deserved. It's all random, & maybe we (even me, often times), should appreciate it more.

 

[etj4Eagle]

Originally posted by Graeme the mad
I was pretty sure you couldnt get a leader from the top of a stack - I always send elite units into cities last because of this.

Where did the leader appear eagle?

It was on desert terrain with one stack of units attacking another stack. This happened when I was doing my combat tests to show that combat was working as given. Guess I should have kept track of leader generation as well. Was actually quite surprised when I got a leader, had forgotten that with fighting all those battles, some of the units would become elite and could produce a leader.

 

[sumthinelse]

I retract my theory about impi not generating leaders. I have a situation where it's impossible to get a leader by killing archers also.

The truth seems to be that *sometimes* the odds of getting a leader is zero no matter if you have 1000 elite victories. If it is possible to get a leader, then the odds are 1/16 or 1/12 with the epic wonder.

 

[TheNiceOne]

Originally posted by sumthinelse
The truth seems to be that *sometimes* the odds of getting a leader is zero no matter if you have 1000 elite victories. If it is possible to get a leader, then the odds are 1/16 or 1/12 with the epic wonder.

You misunderstand the concept about odds. The odds of not getting a leader in 300 elite victories is 1:256 millions. This means that if you repeat (with new random seed) an attack with 300 elite units 512 million times, then you will probably be without any leader two of the 512 times.

This does not mean that the odds was zero 2 of the times (if we assume a real random generator). The odds was, and still is 1/16 for each attack. The result however was zero.

The beauty with random results is that they in fact are random with a wide distribution of results, and not a string of exactly 15 attacks without getting a leader and then getting one the 16th.

What you and others seem to want, is a random generator that does not simulate real random numbers, but one that evenly distributes the results in the (relatively) short run. This may even be a good idea, as it may be very frustrating when you encounter a string of bad luck, but is something completely different than saying that the random generator we have is incorrect.

 

[sumthinelse]

Originally posted by TheNiceOne

You misunderstand the concept about odds. The odds of not getting a leader in 300 elite victories is 1:256 millions. This means that if you repeat (with new random seed) an attack with 300 elite units 512 million times, then you will probably be without any leader two of the 512 times.

Thank you for contribulting to this discussion.

I understand probablilty better than you think.

If what you say (that the odds are 1/256M) were always true, I would not be complaining about the results. The fact is that the odds do not appear to be what you you say they are in many cases. If I do not generate a leader in 200 elite victories the odds are about 400000 to 1. That happened to me, and I would agree that it is possible, but suspicious. And then, in another situation soon after that, in the same game, the only game I have played in 1.21f, another string of 60 elite victories produced no leaders. 48 to 1. In situations in which getting a leader is possible, the odds are 1/16. I can reload the game and repeat generating a leader in these cases. In cases in which it is not possible to get a leader, I could reload the game at that point and try it 200 more times, but I don't think it is worth the effort. You would still tell me that I was having bad luck and I would say you were wrong. If I had a game editor to increase the number of combats per turn in these saved games, I might try it. I do not have confidence that I can recreate these situations by constructing a test scenario, but I might try that some day.

You may not have seen these conditions in your games. If that is the case, good for you. In my game, these situations occur very frequently. I believe it is a bug, or possibly something got corrupted on my installation. I might try reinstalling.

In situations in which it is possible to get a leader, then what you say is true.

In situations in which it is impossible (or nearly impossible) to get a leader, the odds are very different.

I am still playing the game, and I am observing it from a much more objective point of view than you imagine.

 

[Lt. 'Killer' M.]

sumthinelse, I`ve seen it too, instances where I couldn`t get a leader in entire games. Once, I took over the world on a huge map, using only Legionaries, then Knights, then Cavalry. Thousands of battles, and no leader at all.

After a while, I strted to use veterans, and when they got promoted to elite I reloaded and used an elite instead. Should work occasionally - but NO LEADER! NEVER! In at least 1000 elite wins...


Bug? Possibly!
But I guess it`s just my bad luck (I say that so we can save ourselves the 10 post pointing out it`s just bad luck )

I`ve seen things like that repeatedly, especially if I only use 1 kind of troop in my attacks. Combined arms works far better.

Often, I would get leaders. But only 1 nearing the end of the second war, and then another way later on that I`d use for the FP. It was like the game telling me: "You control 4 times the territory of anyone els you don`t need leaders. Oh, now you have 55% corruption so you get one for FP."

 

[TheNiceOne]

Originally posted by sumthinelse

If what you say (that the odds are 1/256M) were always true, I would not be complaining about the results. The fact is that the odds do not appear to be what you you say they are in many cases. If I do not generate a leader in 200 elite victories the odds are about 400000 to 1. That happened to me, and I would agree that it is possible, but suspicious.

I can understand that you become suspicious when something like that happens. But statistically, this should happen once in a while. The question is whether it happens more than it should (which I doubt).

And then, in another situation soon after that, in the same game, the only game I have played in 1.21f, another string of 60 elite victories produced no leaders. 48 to 1.

Its 48 to 1 to get no leaders in exactly 60 wins, but this cannot be seen alone. I'm sure that you got some leaders just before or after (or probably both) the 60 wins.
If you have 64 wins, then the probability of having 60 consecutive wins without any leader increases from 1/48 to 1/38! So by increasing the test with 7% more wins, the probability is increased by 25%. Increase the test to 100 wins (which I guess were in your sample), and you'll find that the probability of this quite much higher than expected.

I'm not stating that there is no bug that temporarily removes possibilities of getting a leader, but I seriously doubt it, because with a good random generator, such strange results shall happen more than seems natural.

 

[Lt. 'Killer' M.]

Originally posted by TheNiceOne



[/B]
Its 48 to 1 to get no leaders in exactly 60 wins, but this cannot be seen alone. I'm sure that you got some leaders just before or after (or probably both) the 60 wins.
If you have 64 wins, then the probability of having 60 consecutive wins without any leader increases from 1/48 to 1/38! So by increasing the test with 7% more wins, the probability is increased by 25%. Increase the test to 100 wins (which I guess were in your sample), and you'll find that the probability of this quite much higher than expected. [/B]

my 2 cents: That is NOT the question! The question is: is there an imbalance that takes away from the enjoyment of playing Civ3 - and there is!

Good RNG or not - it happenes far too often that someone doesn`t get a leader for an extremely long run.
Also, it happenes so often that I get lotsa leaders (like 10 in 50 wins) - which throws all planing out the window (a game like that you can`t loose) that the RNG simply doesn`t fit the game!!!!! It may give a very goos string of random numbers, bell-distribution or not, but the game needs a different one.

Example: the RNG gives numbers for win/loss decision. I try Warrior-Warrior (8HP each) and loose all. Then, reload, try Tank-Warrrior - same thing!

The RNG created 8 numbers at the extreme end of the range in a row! That will occasionally happen, but I see it very often. Thus, for Civ3 purpose, tech advantages are negated very often.

Now what kind of game is Civ3?????? Build-up strategy or numbers?????

 

[TheNiceOne]

Originally posted by Killer


my 2 cents: That is NOT the question! The question is: is there an imbalance that takes away from the enjoyment of playing Civ3 - and there is!

Example: the RNG gives numbers for win/loss decision. I try Warrior-Warrior (8HP each) and loose all. Then, reload, try Tank-Warrrior - same thing!

The RNG created 8 numbers at the extreme end of the range in a row! That will occasionally happen, but I see it very often. Thus, for Civ3 purpose, tech advantages are negated very often.

Now what kind of game is Civ3?????? Build-up strategy or numbers?????

First, it has been proven that strings of numbers at the extreme end of the range happen as often as they statistically should in CIV3. So when you say you see it very often, it is more often than you like, not more often than statistiaclly expected.

But, you have a valid point in that the game could be better if the random numbers were equalized more in the short run, i.e., for each elite win without generating a leader, the chance could increase with 10%, and when getting a leader the probability would be reset.

This would make the game more predictable, but I guess a little bit more predictability could be for the better.

 

[DrJambo]

This thread is daft!

I once got 2 leaders in the same turn, straight after each other. I used the first leader straight away, had another fight and hey presto, another leader was created.

1 in 16 chance does not mean that you get a leader after every 16 battles where an elite unit wins! IT means that everytime an elite unit wins you have a 1 in 16 chance of getting a leader...

1 in 16 is a very small chance, 6.25% to be precise.

Imagine rolling a 16 sided dice and being told you can only have a leader if you roll a 1! Now, from days spent playing RPGs i know this is indeed a very small chance. Instead of thinking how unlucky you are when you don't get a leader, think how extremely lucky you are when you do!!

Arguing about not rolling a 1 on a 16 sided dice after 200 rolls for instance is plain stupid. Just get on with it. Sooner or later one will turn up, but the beauty of it is that maybe 2 will turn up in quick succession! You just never know.

 

[Lt. 'Killer' M.]

Originally posted by TheNiceOne

So when you say you see it very often, it is more often than you like, not more often than statistiaclly expected.

But, you have a valid point in that the game could be better if the random numbers were equalized more in the short run, i.e., for each elite win without generating a leader, the chance could increase with 10%, and when getting a leader the probability would be reset.

This would make the game more predictable, but I guess a little bit more predictability could be for the better.

1) It still seems to me it happenes too often if you compare to living systems. Randomness in nature is checked by evolutionary processes. Man, as a product of these processes, is 'optimized' for them - and thus has his difficulties with 'extreme' randomness. So it depends very much on what kind of randow distribuiton you choose. We expect bell-shaped distribuiton, and we expect to get the strings in the first "Standartabweichung" - don`t knoe in english, must be something like 'standard deviation' (???). We simply don`t expect things like a string of 100,100,100,99,100,99,100,5,100 when drawing numbers from 1-100. We rather expect strings like 45,46,45,45,45,88,45 because in all 'randomness' in real live, extreme numbers are rarer. That`s why many people believe choosing extreme numbers for lottery gives them a lesser winning chance than middle numbers - which explains the odd string of lots of people 'cracking' them same jackpot so often. There uis simply a higher chance for extreme number jackpots not to be cracked, and a higher chance for middle-number jackpots to be split becuase middle numbers are chosen more often.

This is also the way we expect Civ to be, after all it is a simulation of our own history! So, if it doesn`t fit the criteria (as lottery doesn`t) we are surprised by the results.

So i guess we can agree which each other that the RNG as it is now does things that aren`t what the player expects, whatever the reason. Also, I do believe that the is no built-in cheat ofr the RNG. it just may be that total randomness w/o bell distribution favors AI players.

 

[Lt. 'Killer' M.]

Originally posted by =DOCTOR=
This thread is daft!

I once got 2 leaders in the same turn, straight after each other. I used the first leader straight away, had another fight and hey presto, another leader was created.

1 in 16 chance does not mean that you get a leader after every 16 battles where an elite unit wins! IT means that everytime an elite unit wins you have a 1 in 16 chance of getting a leader...

1 in 16 is a very small chance, 6.25% to be precise.

Imagine rolling a 16 sided dice and being told you can only have a leader if you roll a 1! Now, from days spent playing RPGs i know this is indeed a very small chance. Instead of thinking how unlucky you are when you don't get a leader, think how extremely lucky you are when you do!!

Arguing about not rolling a 1 on a 16 sided dice after 200 rolls for instance is plain stupid. Just get on with it. Sooner or later one will turn up, but the beauty of it is that maybe 2 will turn up in quick succession! You just never know.

=doctor=, just have it happen to you and you`ll swear all the time I bet!
It can simply ruin the game (and I don`t mean it will make you loose - it will take the fun out of it!) when you get absolutely NO leader for an eternity, or if you get too many - like 10 in 15 turns or so!

So, I´d say we come back to the age-old idea of increasing the probability of leader generation after a certain dry stretch. See above!

 

[Loopy]

Originally posted by sumthinelse
In situations in which getting a leader is possible, the odds are 1/16. I can reload the game and repeat generating a leader in these cases. In cases in which it is not possible to get a leader, I could reload the game at that point and try it 200 more times, but I don't think it is worth the effort.

TNO has pretty much argued what I'd argue as far as the statistics go so I'll ignore that for now. But I do want to point out that unless you've turned off the "save random seed" (I forget if it's enabled by default), reloading does not change your chances of getting a leader. Games started with earlier patches definately save the random seed and produce the same strings of numbers from load to load.

 

[sumthinelse]

Originally posted by TheNiceOne

I can understand that you become suspicious when something like that happens. But statistically, this should happen once in a while. The question is whether it happens more than it should (which I doubt).

[/B]
Its 48 to 1 to get no leaders in exactly 60 wins, but this cannot be seen alone. I'm sure that you got some leaders just before or after (or probably both) the 60 wins.
If you have 64 wins, then the probability of having 60 consecutive wins without any leader increases from 1/48 to 1/38! So by increasing the test with 7% more wins, the probability is increased by 25%. Increase the test to 100 wins (which I guess were in your sample), and you'll find that the probability of this quite much higher than expected.

I'm not stating that there is no bug that temporarily removes possibilities of getting a leader, but I seriously doubt it, because with a good random generator, such strange results shall happen more than seems natural. [/B]

It did not just happen once. It happens more often than not. I am not talking about something that happened once. It keeps happening. Today I had 170 elite victories without generating a leader, and there were a lot more before that that did not generate a leader. Overall, I guess I have had about 500 elite victories with 7 leaders created. That's about .014, nowhere near 1/16. That would not be so surprising except for the long, extrememly improbable slumps that continue to occur.

Tell you what. I will send you my saved game in one of those turns that I think cannot generate a leader. Before I save it I will fortify all the non-essential units. There will be 3 non-fortified units, 3 elite horsemen. You can load the game and repeat the attacks on the 3 archers as many times as you like. I predict you will give up long before you have 1000 victories without a leader.

"I'm not stating that there is no bug that temporarily removes possibilities of getting a leader, but I seriously doubt it, because with a good random generator, such strange results shall happen more than seems natural."

The infogames support person I contacted might disagree with you there.

I don't have any very serious problems with the combat system, although I have seen some strange results when archers are defending. And I like civ3. But I think that there are cases in which we don't get random results according to the advertised probabilities. What we get are "unpredictable" results, which is not the same thing. That is, we don't know whether the game will apply odds of 1/16 or zero on a given turn.

What about Killer's game in which 1000 elite victories did not produce a single leader? I don't even know the name of the number 10^28 but that should make you stop to think.

 

[sumthinelse]

Originally posted by Loopy


unless you've turned off the "save random seed" (I forget if it's enabled by default), reloading does not change your chances of getting a leader.

I am not playing a scenario. The "save random seed option" is only in scenarios.

Originally posted by Loopy


Games started with earlier patches definately save the random seed and produce the same strings of numbers from load to load.

I think you mean that if you continue to play with an earlier patch you will get the same seed every time you load. If you install 1.21f and load a game saved on an earlier patch you get new random numbers every time you load, at least for combat and in turns in which the probability of getting a leader is nonzero. I suppose it's possible that the game uses new seeds except "sometimes" when an elite wins. I have saved games in which you can retry elite victories, and getting a leader is possible. In those cases, sometimes you get a leader after 3 tries, sometimes 15, sometimes 22, etc. etc.

 

[sumthinelse]

Originally posted by =DOCTOR=


Arguing about not rolling a 1 on a 16 sided dice after 200 rolls for instance is plain stupid.

I am sorry if you think gathering emperical data is stupid. Some of us like to study the game to see what it does. And, no, leaders do not always appear eventually, even after 1000 elite victories.

 

[philippe]

but yes i dont think that a civ had more than 40 great leaders
so its a bit realistic.although its very hard to have one

 

[sumthinelse]

Originally posted by philippe
but yes i dont think that a civ had more than 40 great leaders
so its a bit realistic.although its very hard to have one

philippe,

Do you mean that there is a built in limit in the civ3 software? Or do you mean that no nation in history had 40 great leaders? I you mean the latter, I guess it depends on what you mean by "great leader."

I think the caravans and transports worked a hell of a lot better than the leaders in civ3. At least in civ2 you knew what you were dealing with and could make rational plans based on how quickly you could build wonders.

 

[philippe]

yes like in real i dont think that france had 40 napoleoens
so this rare leader thing is very normal

 

[Mîtiu Ioan]

Originally posted by philippe
yes like in real i dont think that france had 40 napoleoens
so this rare leader thing is very normal

Yes - but France have two Napoleons, one De Gaulle, one Jeanne d'Arc, one Richelieu ... and so on. So probablly almost 10 GL during its history ...

Regards