Combat Odds?

[HoldSteady]

Originally posted by Kuciwalker


It was, it solved your problem completely

Where? Where was that? I missed that bit I guess

Story of my life.. I allways miss the good and helpfull stuff..

[Targonis]

There is some strangeness to it though. There seems to be a random chance that a unit will be "super difficult" to defeat one turn, but if you wait a turn, that same unit will die. This does not use the "random seed on reload", which would make things a bit tougher to verify.

What I did was took 6 units and prepared to attack some enemy units. This was a set of six different axemen attacking 3 enemy axemen. The odds were even, a straight 50 percent chance. Now, for one turn, I would attack them, and lose. Reload, take the next, attack, lose. I did this for all six of my units, and every turn I would end up losing the unit when attacking. Waiting one turn, I would then try and this time I would see the 50 percent that was shown.


This seems to work in both directions, where I could have a 80 percent chance of winning, or 20 percent chance. There are just turns where I lose the fight, even when I attempt to attack with three different units, each of which has an 80 percent chance of winning. It is as if the unit I am attacking has super-luck for that one turn that runs out the following turn.

Obviously in a game with other human beings you can't do this sort of test, but in all of these tests(not against barbarians), it seems like there is a random chance that you could take something like a swordsmen and lose to a warrior. If you have several different swordsmen so it's not just the seed of that unit vs. the warriors if it is multiple swordsmen that can't win on the first attack of the round, can it?

I have seen this over and over again in multiple games, noble difficulty, so it is fairly even in terms of combat odds.

[Kuciwalker]

Originally posted by HoldSteady
Where? Where was that? I missed that bit I guess

Look at the ****ing percentage.

[Kuciwalker]

(xpost with myself)

Then something changed between turns.

Sigh. People, the combat algorithm is known. RNG's are a solved problem as far as civ is concerned. Blake just described it to you, in general terms. Why are there so many people who just don't get that random numbers are completely random and there isn't anything peculiar about how they work in civ?

Sigh.

[HoldSteady]

Originally posted by Kuciwalker
(xpost with myself)

Then something changed between turns.

Sigh. People, the combat algorithm is known. RNG's are a solved problem as far as civ is concerned. Blake just described it to you, in general terms. Why are there so many people who just don't get that random numbers are completely random and there isn't anything peculiar about how they work in civ?

Sigh.

Well, no need to sigh so heavily, it's not as if all your tremendous efforts to explain stuff so clearly were all in vain.. Blake did explain it, prefectly IMO, and I'm sure there must be posts here in which they were solved and explained laboriously, but I guess we missed those..

[NackBarnes]

*laughs* Just about bet those posts weren't from Kuciwalker, eh?

Yeah, Blake explained it very well, very helpful. And, Targonis, I've run into similiar "this guy can't be beat this round" issues myself. Have wondered if "last guy defending the capitol" gets some secret hidden bonus.

[Kuciwalker]

Yeah, Blake explained it very well, very helpful. And, Targonis, I've run into similiar "this guy can't be beat this round" issues myself. Have wondered if "last guy defending the capitol" gets some secret hidden bonus.

This is why I don't like dumb newbie posters, Solver. They all say the same goddamn things that have been refuted over and over. You'll notice I'm nice to the ones who aren't dumb.

Can we have a big topped thread saying something like THE BATTLE ODDS ARE NOT FAKED. THEY ARE CORRECT. THERE ARE NO HIDDEN VARIABLES. QUIT WHINING. ?

[NackBarnes]

This is why I don't like dumb newbie posters, Solver. They all say the same goddamn things that have been refuted over and over.

Ah. Resident Forum Troll. I get it. /ignore

Meanwhile, odds calc is a lot better now in my games, not focusing on the x vs. y statement, now that I have a better idea of what that statement means. Thanks guys!

[Solver]

Originally posted by Kuciwalker
This is why I don't like dumb newbie posters, Solver. They all say the same goddamn things that have been refuted over and over. You'll notice I'm nice to the ones who aren't dumb.

Can we have a big topped thread saying something like THE BATTLE ODDS ARE NOT FAKED. THEY ARE CORRECT. THERE ARE NO HIDDEN VARIABLES. QUIT WHINING. ?

And you'll notice that you're temporarily banned. Again.

I've told you several times... if you can't stand people who ask questions that have already been answered previously, fine. Leave the answering to people who are glad to answer. But then have some common courtesy to keep your mouth shut. I really don't understand why it's so hard to stay out of a thread.

[Kataphraktoi]

What I did was took 6 units and prepared to attack some enemy units. This was a set of six different axemen attacking 3 enemy axemen. The odds were even, a straight 50 percent chance. Now, for one turn, I would attack them, and lose. Reload, take the next, attack, lose. I did this for all six of my units, and every turn I would end up losing the unit when attacking. Waiting one turn, I would then try and this time I would see the 50 percent that was shown.

Im finding this to be true as well.

[Blake]

That's just how a pseudo random number generator (PRNG) works.

Lets say it generates numbers between 0 and 99, and the next 10 numbers in the sequence, are, by chance, all small numbers, like between 0 and 20.

Now it doesn't matter what unit you pick to fight with, it's going to get a bunch of low rolls, and probably lose.

What you need to do is "use up" the bad streak of random numbers, you could for example set a unit to auto-explore (the routine uses up random numbers). One of the most straightforward ways is to just end the turn, so then "someone else" ends up using the bad streak of numbers. Although in reality, the random numbers would just be used up in AI routines other than combat (like auto-explore, or choosing the build), where the numbers wont actually be bad, so don't imagine you can pass a bad streak of combat on to the next civ. But in any case, come next turn, you'll be on a new sequence of random numbers which probably wont be bad (although there is no guarantee because they are random numbers after all).

[Kataphraktoi]

I have\had little knowledge of the games RNG system. What you say ties in nicely with my experiance so far

[Jaybe]

Of course, when you (unknowingly) decide that you are in the middle of an "unfavorable streak of RNGs," it will have just ended, and you will jinx yourself by delaying further attacks.

[Blake]

Indeed, because this ONLY works in the context of Save-Reload, you can never predict what the next RN will be from the previous ones (unless you're using a memory debugger :P) so you can only know them by reloading.

[Kataphraktoi]

Artillery units trump bad RNG rolls