Important question for Jeffrey Morris about combat.

[Rhandom]

Could you please arrange an explanation for the way the game handles its "random" number generator. It is no secret now that the combat results are predetermined, and that reloading the game will yeild the same combat results endlessly, even up to 5 turns in the future (longest I have tested it). It is apparent to many of us that the method of determining "random" numbers leads to long strings of low/high results at a much higher frequency than an actual random number generator, leading to absurd combat results that can ruin games.

I know some people don't see the problem. Good for you. A lot of us have, and its not just our imaginations/whining.

I know some of you think it is an anti reload cheat thing, which is also ridiculous, as there are plenty of other more effective ways of cheating if you want to (just build the map the way you want, for instance), and since there is no multiplayer, who gives a rats ass? Furthermore, if anything it encourages reload "cheating", because you know if an absurdly bad combat occurs, you can wait a turn or use up the "bad" rolls on non essential combatsand definately get a different result.

So please, fanboys, don't turn this into a flame war, just let the people answer. Jeff, could you please explain what is up with the combat calculator/number generator, and why on earth it was selected over a traditional number generator?

Thank you

 

[Rhandom]

In fact, since the numbers seemed to be saved with the game, can we send saved games where long strings of bad combat results occur, so you can see the problem if you currently don't believe there is one?

 

[eyrei]

What exactly is a fanboy? Sounds derogatory to me. If that is the case, why would you expect those you label 'fanboys' not to respond to an insult. I'm really tempted to toss about a few choice insults myself, but I think I'll be content asking how old you are?

 

[treadwin]

Perhaps just combat in general. It may not be the random number generator - Are there other factors we don't know about ??

But do we want to know - is that lack of certainty what makes the game enjoyable ??

(braces self for barrage)

 

[Bugg]

It is no secret now that the combat results are predetermined, and that reloading the game will yeild the same combat results
I don't think this is true, I have lost battles reloaded and won. So it's not that all combat results are predetermined. I think thre might be some kind of "kicker" that gives the A.I. an advantage under certain conditions. For example if A.I. gets more than six techs behind all A.I. units get +12 attack and defense. Then when you attacked a phalanx its like tank attacks at 12, phalanx defends at 14. This way no matter how many times you attacked you would always lose. I'm not saying that this is kicker, but something like this would give game balance to any edited game.

 

[eyrei]

Originally posted by Bugg
It is no secret now that the combat results are predetermined, and that reloading the game will yeild the same combat results I don't think the problem
I don't think this is true, I have lost battles reloaded and won. So it's not that all combat results are predetermined. I think thre might be some kind of "kicker" that gives the A.I. an advantage under certain conditions. For example if A.I. gets more than six techs behind all A.I. units get +12 attack and defense. Then when you attacked a phalanx its like tank attacks at 12, phalanx defends at 14. This way no matter how many times you attacked you would always lose. I'm not saying that this is kicker, but something like this would give game balance to any edited game.

I really don't think it is this bad! When a tank attacks a spearment fortified in a city with walls the formula is something like this: 16/4+16=16/20=80% chance the tank will win each round of combat. There is a reasonable chance that the the spearman wins each round. Not a good chance, but enough that it will happen occasionally.

 

[nklatt]

Originally posted by Bugg
I don't think this is true, I have lost battles reloaded and won.

If you reload and do the exact same moves/battles/etc., the outcome will be the same. If you do something that adds or removes one random factor from the list, the results after that will (have the potential to) be different..

 

[Zachriel]

Generally, computers cannot generate a truly random number. This is due to their intrinsic deterministic structure. There are various algorithms to generate pseudo-random numbers, i.e. numbers which appear random and can usually take the place of random number. Some algorithms are better than others.

The usual way to program the use of random numbers, is to create one, then feed that number back into the algorithm. With this method, the list of random numbers generated will always be the same, when starting from the same initial seed.

Presumably, the numbers at least give equally good and bad values, so your own killer phalanx should win as often as the enemy's killer phalanx -- if the AI is dumb enough to attack a well-entrenched defense without proper planning. Most of the time, I've seen the AI just bypass my strong defenders.

Amazing how much trouble a few riflemen on a hill in Afghanistan are causing to the most powerful, technologically advanced military in history.

 

[Bugg]

quote:
--------------------------------------------------------------------------------
Originally posted by Bugg
I don't think this is true, I have lost battles reloaded and won.
--------------------------------------------------------------------------------



If you reload and do the exact same moves/battles/etc., the outcome will be the same. If you do something that adds or removes one random factor from the list, the results after that will (have the potential to) be different..

Oh I see. Iwas wrong. But I still think it is a game balance kicker, maybe not with the bonus that I gave, maybe something like if an ancient unit is attacked by a modern unit they have one turn of invulnerability. But whatever it is I don't think that they are going to remove it.

 

[Gyre]

I actually like that the results are predetermined. It's just the results themselves that I have a problem with.

 

[Rhandom]

Almost every other game I've played has been able to come with with random numbers, or at least simulate it well enough that we can't tell. This one isn't doing that.

 

[damunzy]

Originally posted by eyrei
What exactly is a fanboy? Sounds derogatory to me. If that is the case, why would you expect those you label 'fanboys' not to respond to an insult. I'm really tempted to toss about a few choice insults myself, but I think I'll be content asking how old you are?

[Modspeak]Thank you for being civil eyeri, but try not to take offense at other posters unless they name you directly. 'Fanboy' is derogatory and the poster may think that you are one, but nothing was said about you so, no harm no foul, this time.[Modspeak]
On topic:
I hope a Firaxisian answers your question. Maybe you would get a better response by emailing them directly? If you do get a response please post it here and PM me.

 

[cutiestar]

Yes well i have lost 3 elite immortals fighting one spearman that was not in a city or on a hill or forest, just standing on the grassland.

But it only worried my strategy, not my opinion of the game or made me think there is a problem that needs changed, it is so rare, it stops me taking the whole combat thing for granted.

I am a *fangirl*

 

[GeniX]

Let us suppose that this game is run on a computer. Right, you with me? Now computers have to calculate things. Its what they do. So we want a random number. It will have to be calculated.

So with a formula, we now get a number sequence which is complex enough that it seems random to an onlooker. This number would be generated from a seed. Sometimes a seed can be the current time - generally its one of the better methods to include the current time in the calculation.

Suppose civ3 just saves the seed with the savegame. No fancy trickery here. Numbers as encountered are random enough.

Now, if what you say is true, then the game is most likely saving the random number seed. Its quite unlikely that it CHOSE you are GOING to lose the next few battles. If I were coding the game, I wouldnt bother coding in some complex system to precalc battles.

So, now we address the issue of 'if the game is saving this seed, then why?'

I used to love playing Civ I. Emporer was very challenging. Sometimes the AI would manage to get to a good city of mine, and I would go 'BUGGER'. What do I do? Save game, next turn. Reload until the AI does not capture my city.

Silly? Yeah, but its still a form of cheating. The idea of a savegame is not that you can replay it until you win, but rather that you save a game at a point to continue it later.

Now, if you werent also trying to 'cheat' by reloading your game until you won, you would not have noticed that the same 'random' numbers seem to be being picked.

If Firaxis did put this in the game, it would probably annoy those who love doing it enough to cause them to start a thread on this message board whining about it.

 

[Rhandom]

Genix, did you even read the original post?

Every other game I own can do random-like number. Except this one. It is not just a matter of the game using a strange seed system at the start of the turn - it uses the same seed system over several turns. Furthermore, whatever system they are using gets long strings of low or high numbers, leading to many bad combat results. And when I say bad combat results, I'm not talking about the rare frigate killing a battleship. I'm talking about 2 spearmen holding back 6 cavalry, unfortified on the flats, or 10 bombing runs that do nothing, and these 1 in 1000 chances happening several turns in row. It is getting a string of not just dumb defeats, but dumb defeats where you don't even score a hit.

As to the anti-cheat nonsense, I repaste the orginal thread. "I know some of you think it is an anti reload cheat thing, which is also ridiculous, as there are plenty of other more effective ways of cheating if you want to (just build the map the way you want, for instance), and since there is no multiplayer, who gives a rats ass? Furthermore, if anything it encourages reload "cheating", because you know if an absurdly bad combat occurs, you can wait a turn or use up the "bad" rolls on non essential combats and definately get a different result."

 

[Rhandom]

Also, Firaxis guys, we would like some sort of explanation, or a place to send save game files if you don't believe there is a problem.

 

[eyrei]

Random does include occasionally strings of low or high 'rolls'. That is the nature of randomness, it is not necessarily even a different number each time. I am guessing the game generates a random number between 1 and 100 and applies this to the units chance of winning that round of combat. If a veteran tank loses to an entrenched spearman, the game simply generated 4 numbers out of 8 or less (assuming the spearmen is veteran) that were 20 or less. This is not ridiculous, it is random.

 

[etj4Eagle]

And a true random number sequence will show some "streakiness" (papers have been written on this subject). This is actually a way you can test the quality of randomness. There are tools out there for detecting fraudulent financial documents that test the quality of the random numbers. Since we humans when asked to generate a sequence of random numbers will generate one that is actually too random, with no or too few improbable results. Hence if the document is "too random" then it was altered.

Now this is not to say that the everything is Kosher with the game. Now if we could log the random number results we might be able to tell if there is a problem with them, in general or only on certain systems.

Also we do need to know the exact probability formula for combat, so that we are not arguing that an event is improbable when it actually is more more probable than we realize.

 

[Rhandom]

Guys, I know how random numbers work. I accept some streakiness. But even the streaks have to obey some sort of probability, or else you don't have random numbers. Getting a "yatzee" in one roll happens. It's pretty damn unlikely to happen happen 3 times in a game. It's even less likely to happen 3 times in a game for 6 games in a row. Thats the kind of screwed up results we are talking about. This isn't about single combats, this is about 5 units losing to one, with 3, 4 or 5 exchanges of blows mandated for each unit, with only a 1/3 chance of losing each exhange.

Again, still waiting for a Firaxis position on the design decision and whether they think there is a problem.Something definative to end the speculation from both sides would be welcomed.

 

[Zachriel]

Characteristics of randomness:

Symmetrical
That is there should be as many winners as losers. Do you ever have your own killer phalanx?

Streaks
There will be streaks of winners or losers of arbitrary length. In other words, there will certainly be 1000000000 winners in a row -- if you play long enough.

Bell Curve
Average streak lengths will follow a bell curve. Most streaks will be short, long streaks rare.

Independence
Each roll of the randomizer is independent. Just because you have 10 heads in a row does not mean the next flip will be heads or the next flip will be tails. They must be independent, or it is not random.

Computers
Computers cannot generate truly random numbers. They can only generate pseudo-random numbers, which resemble random numbers in the features listed above.