Only 1 leader possible from any given elite

[Lt. 'Killer' M.]

Originally posted by etj4Eagle


It is not necessarily a bug that you can't get two GL from a single unit. In some ways I can see the logic behind that decision.

However, what is wrong is that there is no way to tell which units have created their GL and which have not. This non-distinction is the "bug."

OK, , let`s dot the Ts and cross the Ts:

either it is a bug that they made the elite unit incapable of giving more than one leader, or it is a bug that they did so but didn`t make the used iones distinguishable and didn`t tell us, or it is a simple mistake when they didn`t tell us and the non-distinguishability is on purpose

 

[sumthinelse]

Originally posted by Killer

but I didn`t get a single leader in the entire game!!!!!

So the 1 GL per elite does not explain your leaderless game. Are we back to the "extremely bad luck" hypothesis? If you had 1000 elite victories in a row with no leader the odds are 1 in (1 with 28 zeros after it)... and I don't believe you are THAT unlucky. Maybe if you had built more temples

 

[Sir Yelof]

Some have argued that the chance of a GL being spawned is partially dependent on the difficulty of the battle. All anecdotal "evidence", of course, but people have felt that GLs are more likely when you defeat an equal or greater foe, or if your elite unit barely survives the battle.

So I just want to see if there are any holes in your test, or other possible explanations:
- Could the advanced modifications of the tank have made it less likely to spawn a GL? In which case, the early GL was lucky, and the lack of other GLs within normal expectations.
- Could the chance of getting a GL be decreased when a unit has already succeeded? (Of course this would imply that not showing which units have produced GLs is still a bug)

Just a few thoughts...

--Yelof

 

[Lt. 'Killer' M.]

Originally posted by sumthinelse
. Maybe if you had built more temples

starting to like your sense of humor....


seriously, that game was fought with small band of elite units, maybe 10 or 11 of each kind, so if for whatever reason the flag was toggled accidently......

And I do tend to have really long bad stretches and really long good ones, too!

 

[etj4Eagle]

Originally posted by Sir Yelof
Some have argued that the chance of a GL being spawned is partially dependent on the difficulty of the battle. All anecdotal "evidence", of course, but people have felt that GLs are more likely when you defeat an equal or greater foe, or if your elite unit barely survives the battle.

Just for the record my test were modern armor verse spearmen and warriors, just about as undifficult as you could get. And the MA had enhanced stats, so it was even more lopsided victory for me. Yet I was still getting GL at the expected interval. Remember leaders are rare enough that the gambler's fallicy quite easily occurs.

 

[TheNiceOne]

Originally posted by sumthinelse
So the 1 GL per elite does not explain your leaderless game. Are we back to the "extremely bad luck" hypothesis? If you had 1000 elite victories in a row with no leader the odds are 1 in (1 with 28 zeros after it)... and I don't believe you are THAT unlucky. Maybe if you had built more temples

It wasn't as many as 1000 elite victories, was it?
Anyway, even if it was, then think about this:

With 1000 elite victories, there are 2^1000 (1 with 301 zeroes after) different permutaions of results. Those 2^1000 don't have the same probability, the single most likely one is the one with zero leaders, which has the probability you stated. The least likely is one with 1000 leaders, which has odds less than my version of Excel could show
There are 1000 different permutations of getting exactly 1 leader: Getting him after the first combat, after the second, etc. Each of those 1000 permutations have odds 1 in (1.6 with 29 zeroes). Together these 1000 permutations give the odds 1 in (1.6 with 26 zeroes) of getting exactly one leader, slightly better odds than getting zero.

So, when having 1000 elite wins, you will get exactly one of these 2^1000 possible permutations. The one you are most likely to get, is the one with no leaders at all, but since this one is so special, everyone reacts, even though it is the single most likely one.

If I tell you that I had 1000 elite wins and got leaders after the following battles: 17, 40, 102, 104, 110, 192, 216, 231, 387, 412, 498, 512, 526, 698, 700, 715, 743, 759, 781, 792, 854, 870, 901, 911, 919, 971, what would your reaction be: Most likely, you would say that getting 26 out of 1000 rolls was a bit unlucky, but nothing to write about, and that the distribution seems ok.

Well, the fact is that the odds of this permutation is 1 / (4 with 58 zeroes), i.e. a probability very much less than getting no leaders at all.

The "expected" result of 1000/16 = 63 leaders has 1000!/(63!*937!) = 6.8 with 100 zeroes different permutations, but each permutation has a probability of 1 / 1.3 with 102 zeroes. So any single string of result with 63 leaders, have a proabability that is almost zero.

We react when we see the special result of no leaders, but forget that this is actually more likely than any other string of results. It is just that there are so very many other results....

 

[cracker]

Jaybe,

You have these probabilities figure backwards compared to how you should be assessing the leader generation probability.

To provide valid results you need to look at the probability that a leader WILL NOT BE PRODUCED during any given combat sequence.

For a standard combat event (i.e. non military civ non HE present) then the probability of not getting a leader in one combat victory is 15 out of 16 or 93.75%.

You subtract this number from 1 to get the probability that a leader should be produced. So in the case of one combat event there would be just over a 6% chance of getting a leader.

When you have multiple elite combat wins, the probability of NOT GETTING a leader goes down when the results are truly random. This probability is found by raising the single event probability to the power of the number of combat events or 15/16^n

In the case of 11 elite combats, there is about a 50% chance that a Great leader will be produced

16 combats does not guarantee a leader, but only yields a 64% or 2/3rds chance that a leader will be produced.

50 elite victories gives us a 96% chance that at least one great leader will be produced.

100 elite victories is 99.84%

200 elite victories provides sufficient accuracy to assure us that the results of the test would be accurate in 100,000 repetitions of the test for at least five 9's of accuracy.

If you run a test with 200 elite combat victories and do not produce a great leader, it would be safe to say that something is severely wrong. (either RNG is faulty or leader generation program is blocked deliberately or faulty).

 

[sumthinelse]

Originally posted by TheNiceOne


The one you are most likely to get, is the one with no leaders at all, but since this one is so special, everyone reacts, even though it is the single most likely one.

We react when we see the special result of no leaders, but forget that this is actually more likely than any other string of results. It is just that there are so very many other results....

To paraphrase Will Rogers: There are lies, damn lies, and permutations.

Yes, this is a clever way to look at things, and I believe that you are posting this in a light-hearted spirit, to make us think. And I welcome your comments.

But you do agree that getting *at least one* leader in 1000 tries is almost certain?

Yes "we react" when we see the game apparently doing strange, and extremely improbable, not just a little improbable, things. And sometimes we discover something wrong with the way the game works. My view is that the "one leader per elite" was a problem until we knew what was wrong, and even now, when we know how it works, it still affects playability.

 

[cracker]

For the record, if someone at Firaxis confesses to the existence of this one leader per elite flag or if one of our deep code hackers can reveal the existence of the flag independently, then there should be some sort of visual marker to indicate when the elite unit has earned this honor.

Probably the easiest way to do this would be to change the shape of the little doohickey that sits on top the health bar so that an elite unit that already has won the the Great Leader honor would have a small cube, cone, or pyramid instead of the standard ball shape. It would be nice if the right click menu also had an asterisk of some other indicator that would help identify the elite units that have already been decorated.

It would also be nice to know if the flag stays set when an elite unit is upgraded. We know the uprade resets the elite units back to veteran, but if the GL suppression flag stays set, then this might further impact unit promotion and GL production even at the next unit upgrade level.

These issues are important not just beacuse it is evil and deceitful to hardcode something that makes it impossible for an event to occur without providing some documentation, but also because these issue effect strategy in a big way. Instead of preserving elite units as if they are treasure, once the produce a GL we may wish to disband them or upgrade them quickly depending on which of the sets of conditions apply.

 

[sumthinelse]

Originally posted by cracker

It would also be nice to know if the flag stays set when an elite unit is upgraded. We know the uprade resets the elite units back to veteran, but if the GL suppression flag stays set, then this might further impact unit promotion and GL production even at the next unit upgrade level.

Interesting question. I'll try to check if an upgrade resets the flag when I get a chance.

 

[etj4Eagle]

Originally posted by cracker
It would also be nice to know if the flag stays set when an elite unit is upgraded. We know the uprade resets the elite units back to veteran, but if the GL suppression flag stays set, then this might further impact unit promotion and GL production even at the next unit upgrade level.

Well Cracker just undertook your question, and the flag is reset when a unit is upgraded. I used the same methodology as the second test, except this time the units were tanks. Once I produced a leader I used him to rush a barracks and then upgraded the tank that generated the leader into modern armor.

I then proceeded to produce a leader with one of the other tanks. Then after reloading produced a leader is the same number of turns with the other tank. And then finally reloaded with the modern armor and again produced a leader again in the same number of turns.

Therefore: Upgrading a unit clears the "produced a leader" flag." Allowing you to get another leader once you have upgraded the unit.

 

[sumthinelse]

Originally posted by etj4Eagle


Well Cracker just undertook your question....

Should say "Cracker, I just undertook..." You did, the research, not Cracker, right? But Cracker's question was a damn good one.

etj4Eagle, thank you again for your latest research. This gives us *some* hope of minimizing (but not eliminating) the confusion. This will be useful *if* we have a single "used" elite unit (not in a stack with identical-looking "unused" elite units) *and* there is a barracks somewhere that is not too difficult to get to, *and* there is an upgrade available.

Now I can dedicate my time to another of mankind's greatest problems, whatever that turns out to be.

 

[God]

Wow, in my current game as Japan, monarch, huge map, so far I have got 10 leaders, have rushed all the Middle Age wonders, except Copernicus, which I was beaten too, as my leader approached to build it.

 

[SirPleb]

etj4Eagle, thank you so much for all the testing you've been doing and all the info you've posted! Really good stuff!

 

[TheNiceOne]

Originally posted by sumthinelse

To paraphrase Will Rogers: There are lies, damn lies, and permutations.

That one was new to me, I've always heard that third was statistics...

Yes, this is a clever way to look at things, and I believe that you are posting this in a light-hearted spirit, to make us think. And I welcome your comments.

But you do agree that getting *at least one* leader in 1000 tries is almost certain?

Yes, the posting was certainly in a light-hearted spirit, but there is still some truth int it. The results that look weird to humans are just as likely as any other single result.
But I do of course agree that getting at least one leader in 1000 tries is almost certain (although I was tempted to say that since there are two possibilities: either you get one or you don't, the odds are 1:2 )

Yes "we react" when we see the game apparently doing strange, and extremely improbable, not just a little improbable, things. And sometimes we discover something wrong with the way the game works. My view is that the "one leader per elite" was a problem until we knew what was wrong, and even now, when we know how it works, it still affects playability.

Agreed.

 

[sumthinelse]

Originally posted by TheNiceOne

That one was new to me, I've always heard that third was statistics...

Correct. The third one in the original Will Rogers quote was statistics. When you "paraphrase" something you don't quote it exactly. At least that's the first definition in the dictionary.

Did you ever see the movie "Rosencrantz and Guildenstern are Dead" ? In it, one of the guys keeps flipping a coin and it comes up heads 30 (or maybe 50 or 100, I'm not sure) times in a row. The other guy asks, "Do you know what the odds of that are?" The coin flipper replies, "Yes. It's 1/2." Which is true if your scope of consideration is only one event.

 

[Mike B.]

Each unit can produce at most one great leader. Each time a unit is upgraded, however, it can once again produce a great leader. You can have more than one great leader per game but only one at a time.

 

[Catt]

Mike B: Thanks for the facts!

etj4Eagle: Thanks for the research and work to expand our understanding of the game mechanics!

 

[Sanaz]

Originally posted by Mike B. FIRAXIS
Each unit can produce at most one great leader. Each time a unit is upgraded, however, it can once again produce a great leader. You can have more than one great leader per game but only one at a time.

Thanks for the confirmation! So I think the obvious question is, can we have a change so that the Elite Unit which has created a Great Leader be somehow marked as such? If we have a stack with 4 identical elite units, we currently have no way of tracking which one actually created the leader. Some suggestions were to give this Elite unit extra hit points, extra strength, or extra movement, or maybe a different symbol at the top of the current hit points, or maybe a physical change in the appearance of the unit itself (like a medal of honor). The way it is now, it is very difficult to tell the difference (especially if there are 12 Pikemen defending a city, many of them Elite - we may not even know which one was attacked).

Thanks for reading these lists and responding!!

 

[Mike B.]

Originally posted by Sanaz

So I think the obvious question is, can we have a change so that the Elite Unit which has created a Great Leader be somehow marked as such?

I'll look into it.