How Come...

[NickyH]

Omg....lost 5 modern armor's to spearman once....cannot describe the anger....

It would be nice if someone with math skills could calculate the odds for that.

A couple of days ago a friend of mine got five military leaders in one turn. The probability for that isn't great, but I guess your modern armors are even far more unlikely.

 

[Sharwood]

It would be nice if someone with math skills could calculate the odds for that.

A couple of days ago a friend of mine got five military leaders in one turn. The probability for that isn't great, but I guess your modern armors are even far more unlikely.

I've gone whole games without getting five military leaders. Fighting wars constantly. Arrrrrrgh!

 

[Turner]

Does anyone here who talks about the "RNG" actually care to do the probability calculations?

No, not really. Not me, at least.

 

[Aabraxan]

Not only that, it seems to store the outcome of battle in order, not by units. Like, if you reload and do exacly the same moves, you'll get the same results. But if you change the order you made your moves, the outcomes might change too.

The hidden variable that is provided by the RNG will still be the same though. If there is a difference in the outcome, it'll have to do with a difference in the 'visible' variables (such as attack strength, defense, terrain bonus etc).

I have a question about this, Lord Emsworth. . . . or for whoever can answer it. The RNG governs SGLs, too, right? A while back, I decided that it'd be fun to play one of those HOF-style, Russian space race games using a pangaea map, lots of scouts and lots of scientific civs. At one point, I was 1 turn from a tech and saved my game at the beginning of the turn. I played the turn, hit enter, and got an SGL. For some reason, I didn't save the game before I shut down. When I went back to the game, I thought, "I'll just load up that last turn and grab that SGL." Nope. I did not get the SGL the second time around. Because it was a solo game, I didn't feel bad about reloading. It took me a number of tries, but I was eventually able to generate an SGL again. Yes, I had Preserve Random Seed on. In the case of SGLs, they're not governed by a/d/m, terrain, etc. What makes the difference in their case? Why was it that I generated the SGL the first time, but not other reloads of the same turn?

 

[Lord Emsworth]

I have a question about this, Lord Emsworth. . . . or for whoever can answer it. The RNG governs SGLs, too, right? A while back, I decided that it'd be fun to play one of those HOF-style, Russian space race games using a pangaea map, lots of scouts and lots of scientific civs. At one point, I was 1 turn from a tech and saved my game at the beginning of the turn. I played the turn, hit enter, and got an SGL. For some reason, I didn't save the game before I shut down. When I went back to the game, I thought, "I'll just load up that last turn and grab that SGL." Nope. I did not get the SGL the second time around. Because it was a solo game, I didn't feel bad about reloading. It took me a number of tries, but I was eventually able to generate an SGL again. Yes, I had Preserve Random Seed on. In the case of SGLs, they're not governed by a/d/m, terrain, etc. What makes the difference in their case? Why was it that I generated the SGL the first time, but not other reloads of the same turn?

I simply don't know about SGLs at all.

I think that there are several things/events that use up the RNG numbers that had been generated in advance. Which these things/events are exactly I don't know, but battle rounds are amongst them definitely, and if I am not mistaken certain things concerning diplomacy as well. There might be others. So, it is simply possible that you played your turn slightly different and in doing so eventually got a different RNG number that factored in to whether SGL/noSGL.

On the other hand it might equally well be possible that wrt SGLs it all works a little different, and that the RNG number is created on the fly and has nothing to do with the numbers that factor in to battles etc, or that ... .

(And anyway, I do not have any kind of intimate, quasi-revealed knowlegde of how game mechanics work. Most of that only comes from observation, either my own or that of other people. So, your conclusions are as good as mine. )

 

[ecuwins]

^, ^^ Re-read Rodrig0's post. The same thing applies to SGLs.

 

[Rodrig0]

The hidden variable that is provided by the RNG will still be the same though. If there is a difference in the outcome, it'll have to do with a difference in the 'visible' variables (such as attack strength, defense, terrain bonus etc).

Finally, I got the proof of what I said!
And it's here... A saved game, Conquest, version 1.22, no mods.

1) Wake the elite horseman near Oea, then attack the celtic archer near the city. The horseman will win without a scratch. Now wake a numidian merc from Theveste and attack the bowman near the worker. The numidian will lose.

2) Wake the numidian merc from Theveste and attack the bowman. The numidian will win. Then, wake the horseman mentioned above and attack the celtic archer. The horseman will lose.

So given the exact same units, same terrain, why weren't the results the same? As I said, I think what is stored is the result of the battles, in order, not by units.

 

[Mathias]

First of all, the RNG doesn't directly determine the result of the battle. Rather, it determines the result of each round of battle, in which one unit will lose one hit point. The variable store is a single number, the value of which determines who wins the round. Each time the number is used, the next number is calculated using a mathematical formula and the value of the previous number. This generates a repeatable series of predetermined values based on the initial seed value.

In the scenario you presented, the first case uses eight rounds to determine the outcome of the two battles. The second case needs eleven rounds to determine the outcome. It can be shown that the results of the first eight rounds of the second case are consistant with the eight rounds in the first case. In both cases, the first battle used four rounds. Three of those rounds favored the attacker, resulting in a loss for the defender. The rounds of the second battle favored the defender, resulting in a loss for the attacker.

So, while it appears that the battles may have been determined to be 'win-loss', it is actually the individual rounds of those battles that are predetermined. The strengths of the units involved in this particular situation were close enough that the end result was the same.

Edit:
Extend this scenario with a third battle by attacking the celtic archers with a horse from Oea. Following the first case, that horse will lose. Following the second case, that horse will win.

 

[Lord Emsworth]

Finally, I got the proof of what I said!
And it's here... A saved game, Conquest, version 1.22, no mods.

1) Wake the elite horseman near Oea, then attack the celtic archer near the city. The horseman will win without a scratch. Now wake a numidian merc from Theveste and attack the bowman near the worker. The numidian will lose.

2) Wake the numidian merc from Theveste and attack the bowman. The numidian will win. Then, wake the horseman mentioned above and attack the celtic archer. The horseman will lose.

So given the exact same units, same terrain, why weren't the results the same? As I said, I think what is stored is the result of the battles, in order, not by units.

I can't look at the save, but I don't think I actually need to. In any case though, you are absolutely right. The rng numbers that determine the outcome of the battle-rounds are (a) stored in advance and (b) independently of the individual units. But not independently of the individual battle rounds.

But that is exactly the reason, why there is no, and cannot be any, cheating on the AI/PRNG side wrt elite units etc. The PRNG would have to have some sort of foreknowledge of what the player is going to do. Will you take the vet, first and then the elite, or will you do it the other way 'round? The computer would need to know that, if there were some cheating.


ETA: But don't forget that 'retreat' might need to be determined as well in the case of fast units, which uses up RNG numbers.