Remarkably predictable random combat....

[aahz_capone]

I've noticed it too, and as has been said, there are good sides and bad sides. Even after I noticed it, I've stopped "cheating" months ago as well. I can also say that unlike in civ2, when the goody huts WERE totally random (just save before and keep going until you get a science), here the huts have a value for every 2 or three turns, then goes to something else. (or so I gather)


Anyway, I saw that comment about the tugboat and I must agree. even though I don't get (m)any of the fabled "tank vs spearman" story, I DO ALWAYS have my battleships destroyed by rubber dingies. I don't care if you Admiral Nelson, an Ironclad at full would be squashed by a WW2 battleship at 15 miles.

[Grumbold]

I'd like to support Aeson in his analysis of how the random generator works. If you try attacking with a cavalryman and get spanked, reload and try it with a tank instead then you may still lose or the different A/D ratio may allow you to win.

You fight so many combat rounds in an average game of Civ that getting one or two really unlucky streaks per game is virtually inevitable. You only really notice them when the odds are greatly in your favour. If you absolutely cannot stand to lose a good unit then reload and use up those bad numbers bombarding terrain or something.

[Ka Shima]

Two words:
SMAC Ironman
'nuff said

[Grumbold]

SMAC is not a word

[aahz_capone]

SMAC...

Orgasm in complexities...

All hail SMACaholics! Ironmen unite!

SMAC40:68posts

[ACooper]

Originally posted by Aeson


I don't think you understand how the game decides battles. There is a random number generator, the whole point of an RNG is to have varied results. Without one, there would be no chance in the game, the highest A/D value (with terrain/fortification modifications) would always win. The seed is saved in the save file. When the save game is loaded, it puts the RNG back in the same state as it would have been if you played straight through. It doesn't change the A/D values of your units or give the AI an advantage.

The only thing it prevents is the player constantly reloading to always get "good" numbers out of the RNG. If that was allowed, there wouldn't be a need for an RNG in the first place. Every result would just be what the player wanted it to be. If that's what you want, just open up HighScores.cv3 in a text editor, and type in whatever score you want. I personally don't think it makes for a very interesting game though.

This is exactly right. One thing to keep in mind is that a battle in Civ 3 is kind of like a boxing match. (Turn based like the rest of the game.) The two combatants trade blows until one is out of hit points.

I think the major problem is that people don't understand random numbers and computers. Here is a good basic description of how it works.

Please get some real information before ranting on how the computer is cheating against you.

[ShuShu]

To be fair...

... I don't think it is accurate to acuse a thread starter of hijacking his own thread. (I believe the liberal use of WTF makes the 'his' a safe assumption...)

Buried deep within his initial post: "...Different Order...".

If his observations are accurate, then he has got a vary interesting point.

Unfortunately, even if he did do as he said, there is still a statistical probability that two completely diferent series of random events result in the same result.

The most obvious would be: Say he did 10 attacks and they were all Tank vs. spearmen on the same terrain. Changing the order of attack would not change the results because the attacks were all isomorphic.

I realise, that Firaxis bashers are not the most rigorous in their rants, but let us at least assume there was some variation in his attacks.

Suppose the first shock was a tank losing to Megalopolis fortified Elite Spearmean on a hill over the river. The tank loses. The second attack was a tank against the spearman on open ground, and this tank loses as well. When we reverse the order, the second attack will lose because it was a loser for much worse odds the first time. The first Attack may still lose because the dice were worse than needed to defeat the first tank. The appearance would be that the results were pre-determined, when the reality was a combination of two unlucky battles that would always have resulted in losses no matter what the order.

To conclude, while I feel it fair to ridicule our colleague for lack of rigor when providing numbers to support his hypothesis, I don't feel it is fair to accuse him of not knowing what he's talking about until he shows us the the numbers to back it up.

Of course, until then, its just another UFO sighting. (Non-reproducable observation saying more about the observer than the observed...)

[The Andy-Man]

a)Not every one reloads (i didnt) just like not every one played ICS.

b)It should have been a togglerable option

c)Some Civ2 people used the reload to work out stuff for every one else, aswel as the built in cheat mode. they would use these and TEST stuff so we could understand more about the game. So it is not just an aid to cheat.

[Technoron]

Please get some real information before ranting on how the computer is cheating against you.

*sigh* I'm a sunny day programmer. I understand that the computer does what it is told to do - even if I think I told it to do something else (which is usually the case)... I realize that computers are incapable of cheating. In any case, I haven't been ranting about cheating. At no point have I accused the computer of cheating. I've been ranting about how things seemed to be predetermined. Please get some real information before attacking me on a personal level. K?

I think the major problem is that people don't understand random numbers and computers. Here is a good basic description of how it works.

Neeto site. Thanks for the link. So pseudo-random generators do NOT generate a series of random numbers. They generate a predictable string of values based on the initial seed. Hence the use of seeds creates the "remarkable predictability" I was wondering about.

I don't think you understand how the game decides battles. There is a random number generator, the whole point of an RNG is to have varied results. Without one, there would be no chance in the game, the highest A/D value (with terrain/fortification modifications) would always win. The seed is saved in the save file. When the save game is loaded, it puts the RNG back in the same state as it would have been if you played straight through. It doesn't change the A/D values of your units or give the AI an advantage.

I'm beginning to wonder if my posts are being read differently than they were intended to be read. *ponder*. Or is it the other way around... Anyway, I never implied that unit strengths were being changed by the RNG and I know I never wrote anything about the computer getting an advantage (read cheating).

Yes, I understand that there is a RNG. Yes, I understand that the whole point of a RNG is to produce varied results. Thing is, I'm not seeing varied results. Niether are YOU. Nobody is. The seeds make things happen in a predetermined way - thereby removing the "chance".

There's (at least) a couple ways to look at this. One side of the table says that the seeds themselves are randomly generated and thus chance is preserved. My side of the table doesn't think chance is being preserved if current events are based on previously determined seed values.

that's really not what the thread is about. that is a slight threadjack. modern units when most of the time anyways... So could say that power still does mean something, just not as much as it used to.

*ponderponder* errr. yeah. Guess I jacked my thread. But only a little. And not for very long. Thanks for keeping me honest (ya damn nitpicker ).

I realise, that Firaxis bashers are not the most rigorous in their rants, but let us at least assume there was some variation in his attacks. .....To conclude, while I feel it fair to ridicule our colleague for lack of rigor when providing numbers to support his hypothesis, I don't feel it is fair to accuse him of not knowing what he's talking about until he shows us the the numbers to back it up.

Mmmmmmm. Subtle. Very nice.

Of course, until then, its just another UFO sighting. (Non-reproducable observation saying more about the observer than the observed...)

Non-reproducable observation my butt. We've both got the same code. Fire it up and reproduce it for yourself. Seeing is believing.

Unfortunately, even if he did do as he said, there is still a statistical probability that two completely diferent series of random events result in the same result....The appearance would be that the results were pre-determined, when the reality was a combination of two unlucky battles that would always have resulted in losses no matter what the order.

How many times would you need to smell the wet dog before you become convinced that there is, in fact, a wet dog nearby?

Buried deep within his initial post: "...Different Order...".

mmmmm. k. I just did an experiment with a bunch of ancient units and no longer have any clue how combat is resolved.

1. If you execute the same combats in the same order you get the same results every time.

2 If you change the order of the combats but the A/D/M values stay the same, you get the same results in the same order applied to different units.

3 if you change the A/D/M values for the first units to engage and keep subsequent battles exactly as they were, the results for ALL battles will now be different.


Conclusion:
1. Combat results are not predetermined as I had said previously, since I can alter results by changing match-ups and order. However, combat is also FAR from being random. There's a big, ugly formula running the show, and one combat seems to have impact on the next.

2. If you want to "beat the seed" it's a friggin snap to do, provided you have some diversity in the unit A/D/Ms.

3. I have a difficult time believing the seeds were intended to deter save/restore.

I wonder how many numbers get saved?

I just did another experiment.

1. Game turn X
I popped 7 huts on the same round with 7 scouts. Huts were popped in different (but no particular) order three different times. All three times I got the same stuff in the same order. Then I ran the same experiment on the following game turn.

2. Game turn Y
this time I got the same stuff in the same order for all seven huts all three times. But the stuff was different from the stuff in turn X.

3. Game turn Z
same stuff from all 7 huts all three times. Different than the stuff in both turns X and Y

4. Closed out civ3. Restarted civ3. Reloaded Game turn X
Same stuff in the same order for all 7 huts all three times. Same stuff as in the original runs of turn X.

5. Reloaded game turn X. Popped two huts and got the results I was expecting. Engaged in combat. Popped the next five huts and got different stuff than the huts had previously produced.

6. Repeated #5 above. Everything, including the combat result, turned out identical..

Conclusion:
1. My head hurts. and this post is too friggin long.

2. there is only one seed. Period. The whole damn game operates off of a single random seed number generated at the beginning of each turn. Any time an event requiring "randomness" occurs, the computer plugs that seed number into the pseudo-random number formula to get the next value in the series. That value is then plugged into a formula or table or both to determine the outcome.

3. Randomness, or chance, in the game is a composite of the seed and the order of the player's actions. The player, quite literally, creates his/her own luck.

***
Ok folks. That's where I'm at up to this point. If you think I'm wrong, and I may very well be (I'm just a sunny day programmer), say so. If you have your own opinion of how this thing works I'd like to know about it.
***


I KNOW I'm seeing the situation perfectly clearly. If my glasses happen to be smudged...well...

[Whoha]

Since it can not be proven that the ai does not know the seed you will ultimately fail in your attempt to convince those who believe it does.

[Aeson]

However, combat is also FAR from being random. There's a big, ugly formula running the show, and one combat seems to have impact on the next.

There is no way to get a computer to generate a truely random number. Psuedo random numbers can be generated by taking seed values from a constantly modified source (like the system clock) and running them through a formula to produce the "random" result. It seems random because there isn't any visable mechanism determining the numbers. The results are still deterministic, it is just difficult to see how the users interaction with the system was affecting it.

In Civ 3, the initial seed for each game is also determined randomly. Otherwise all maps with like settings would be the exact same, and poping that 4000BC hut next to where you start would always give the same result. In the editor you can generate maps with a predetermined seed to see this work.

Each time a random number is needed, the RNG produces a number based on the number that came before it. When a game is saved, the RNG state is also saved. Reloading just keeps the numbers the same while reloading as they would have been if the game was played straight through. This allows the programmer to create a RNG which gives what they feel is a balanced distribution of numbers.

Neeto site. Thanks for the link. So pseudo-random generators do NOT generate a series of random numbers. They generate a predictable string of values based on the initial seed. Hence the use of seeds creates the "remarkable predictability" I was wondering about.

Sounds like you understand what is going on pretty well now. Does it still bother you that the game works this way? No matter how they set it up, it will always be one "big, ugly formula" running the show, some will just be a little more transparent than others.

[Dis]

whoops, my bad

Amazingly this really pissed my off at first. But really it is only a minor complaint I have about the game. There are many more things that are wrong that need fixin'

but what problems I do have with combat, I fix with the editor. Starting especially with the navy.

yeah ironman would have been a nice feature to include in this game. I can't figure out why they did the things they did in this game. It's like they are forcing us to endure the game the way they want us to play it. I'm suprised they even included an editor .

[Ironikinit]

I've mentioned this before, but Colonization had the predetermined numbers thing. It's exploitable, actually. Through trial and error it's possible to play turns out so that they go as well as possible, maybe not as well as one could cheat Civ II into, but pretty well. Of course, that takes more patience than possessed by most players, esp. with the long reload times. But hey, if you're really bound and determinded to reload, you can do it. We don't know their motives for including this feature, it might not be intended so much to prevent cheating as to... I dunno, maybe it was easier to program.

Anyway, if you want to cheat, cheat. Go right ahead, just I'd rather not hear complaints about it. There are so many better ways to cheat. Why not doctor up a map so it has a continent of nothing but bonus tiles? Why not make the jaguar warrior 15/15/15? Please stop being trifling. Go big or stay home, right?

As for iron man, if one wishes to play it, just turn off autosave. Isn't that about all ironman was? Well, it made the game shut down when you saved and gave bonus points for playing ironman, but points aren't really that big a deal with Civ 3, IMO.

[Zachriel]

Think of the seeded results as a pre-shuffled deck of cards. The results are pre-determined, but unknown. That makes them random for the purposes of the game.

[Thrawn05]

They did that so SOME people don't reload the game every turn if they lose a unit. Same goes with planting spies. The AI also follows the same city planting pattern on the same map no matter how many times you start from scratch again.