Ive noticed that often times during a combat sequence, one side will win consecutively three or four times in a row. Which is really annoying when the one winning so much should be loosing. Becasue of this tendancy, I still think it is possible for the underdog to score damage. However, what would normaly appear as a "streak" would only cause one point of damage. This system will be more profound when the differences between the two units are great. Odds are, with the new system a spearman will never defeat a tank. Unfortunately, it also seems that the time of calvary attacking infantry fortified in large cities might be over (unless carried out with even larger amounts of calvary). However, when two units have similar values the relusts might not actually differ much from what they did before. All in all I think it just means that you wont be able to win using antiquated stuff. I believe it will make the game better. Historicaly large tech gaps usualy lead to one side getting trounced, such as all the natives in North and South America did. I think it kinda goes back to the civ II days of firepower without actually using the firepower system. I guess we we soon see.
Civ III: Conquests Patch Notice
- Thread starter Tavis
- Start date
I'm not sure it's accurate to say that the combat would be less "streaky" as much as to say that there will be less freak chance. Streaks are still statistically possible and equally likely as before. But extreme results are much less likely.
(I suppose this is to say that you can still end up with streaks in which 5 swordsman kill 5 spearmen in a row, but very rarely will you see one spearman defeat one tank).
(I suppose this is to say that you can still end up with streaks in which 5 swordsman kill 5 spearmen in a row, but very rarely will you see one spearman defeat one tank).
Aggie
Deity
Good news!!
Obviously a patch is needed more than ever. Atm the game is unplayable
.
Obviously a patch is needed more than ever. Atm the game is unplayable
.I cant seem to let this concept go without further exploration. Playshogi mentioned that if a unit had 60% odds of winning, it would now actually have a greater than 60% odds. This is absolutely correct. I used this method to determine actual odds.
There are four possible out comes for each series of four rolls (W = win, L = loss)
W,W,W,W
W,W,W,L
W,L,L,L
L,L,L,L
W,W,L,L is not considered because as stated, this result would be re rolled.
Suppose a Win probability of 60%.
.6 x .6 x .6 x .6 = 0.1296
.6 x .6 x .6 x .4 = 0.0864
.6 x .4 x .4 x .4 = 0.0384
.4 x .4 x .4 x. 4 = 0.0256
Add these four numbers and get 0.28. This value represents all possibilities.
The total for the two possibilities that result in an overall win is 0.216
The total for the two possibilities that result in an overall loss is 0.064
Then, using the same principle that 1 divided by six gives the probability of any side of a die appearing on top.
0.216 divided by 0.28 = 0.771 or approximately 77%
and therefore the loss percentage must be 23%
So 60% is transformed to 77%.
Using this method, the actual odds are exaggerated the higher the base odds get (if that makes any sense).
For example, lets assume a spearman whose defense value is calculated at 4 after all factors are considered (this is just for argument) is attacked by a tank of attack value 16. With these numbers, the tank should have an 80% base chance of victory. Using the method above this would actually result in approximately a 98% chance of victory.
In my defense, I can not be absolutely sure this is how things will actually happen. But I am pretty sure these statistics are sound. Please correct me if I am wrong.
There are four possible out comes for each series of four rolls (W = win, L = loss)
W,W,W,W
W,W,W,L
W,L,L,L
L,L,L,L
W,W,L,L is not considered because as stated, this result would be re rolled.
Suppose a Win probability of 60%.
.6 x .6 x .6 x .6 = 0.1296
.6 x .6 x .6 x .4 = 0.0864
.6 x .4 x .4 x .4 = 0.0384
.4 x .4 x .4 x. 4 = 0.0256
Add these four numbers and get 0.28. This value represents all possibilities.
The total for the two possibilities that result in an overall win is 0.216
The total for the two possibilities that result in an overall loss is 0.064
Then, using the same principle that 1 divided by six gives the probability of any side of a die appearing on top.
0.216 divided by 0.28 = 0.771 or approximately 77%
and therefore the loss percentage must be 23%
So 60% is transformed to 77%.
Using this method, the actual odds are exaggerated the higher the base odds get (if that makes any sense).
For example, lets assume a spearman whose defense value is calculated at 4 after all factors are considered (this is just for argument) is attacked by a tank of attack value 16. With these numbers, the tank should have an 80% base chance of victory. Using the method above this would actually result in approximately a 98% chance of victory.
In my defense, I can not be absolutely sure this is how things will actually happen. But I am pretty sure these statistics are sound. Please correct me if I am wrong.
mjmillar
Chieftain
I agree, this will remove the chances of units getting "lucky" and will mean that an attacker with any kind of advantage will almost always win, and vica-versa. I recognise that this is a beta release, but this may need to be studied carefully, as I feel it may remove much of the 'random' element of individual battles.
SirPleb
Shaken, not stirred.
Thank you Jesse! It is very welcome news that Firaxis is supporting us at this level. I think that many of us will be very happy to have a patch which attempts to address the most serious issues, even if it isn't the final supported version. (Which I don't think anyone would expect so quickly!)
One specific issue I want to comment on is the modified combat calculation.
Clearly the new system will effectively increase the difference between attacker strength and defender strength. I've done some quick programming to see the impact of this. I have a quick and dirty calculator now which I think will show the difference in results for old way vs. the new way for any given attacker/defender combination.
If my calculator is correct then I think the results are disturbing in how game-altering they are. For example, consider a veteran Swordsman attacking an unfortified veteran Spearman on grassland.
With the current combat calculation the Swordsman wins almost exactly 2/3 of the time, the Spearman 1/3 of the time.
With the new calculation, if my code is correct, the Swordsman will win 84.4% of the time, the Spearman 15.6% of the time.
In other words, the Spearman's chance of surviving has been cut to less than half of what it was, from 1/3 to about 1/7.
Has Firaxis done this kind of calculation internally for typical scenarios such as Swordsman vs. Spearman? If so then it should be easy to quickly check my calculations. If they are incorrect could you post the right values for this example? And if my calculations are correct, do you really want to make such a hugely game altering change at this late date?
[Edit: To show another example, my calculations show that the odds for a veteran Immortal or veteran Knight vs a veteran Spearman, unfortified on grassland, will change from 79.2%/20.8% to 97.2%/2.8%.
Game-altering indeed. Persia will be unstoppable
]
@derekroth: I think you are on a right track to working out the new odds except for one complication which must be added: you need to work out and accumulate the odds for each separate sequence which results in an overall win or loss, i.e. must allow for W,W,L,W, W,L,W,W, etc. in the overall weighting of results. If you don't do that, the values for the W,W,W,W and L,L,L,L sequences influence the result out of the correct proportion.
And you also need to allow for the number of HP involved. Your calculation of 80%/20% for the tank is true only if both units have just 1HP. If they both have 3HP the result would change to 94.2%/5.8%.
One specific issue I want to comment on is the modified combat calculation.
Clearly the new system will effectively increase the difference between attacker strength and defender strength. I've done some quick programming to see the impact of this. I have a quick and dirty calculator now which I think will show the difference in results for old way vs. the new way for any given attacker/defender combination.
If my calculator is correct then I think the results are disturbing in how game-altering they are. For example, consider a veteran Swordsman attacking an unfortified veteran Spearman on grassland.
With the current combat calculation the Swordsman wins almost exactly 2/3 of the time, the Spearman 1/3 of the time.
With the new calculation, if my code is correct, the Swordsman will win 84.4% of the time, the Spearman 15.6% of the time.
In other words, the Spearman's chance of surviving has been cut to less than half of what it was, from 1/3 to about 1/7.
Has Firaxis done this kind of calculation internally for typical scenarios such as Swordsman vs. Spearman? If so then it should be easy to quickly check my calculations. If they are incorrect could you post the right values for this example? And if my calculations are correct, do you really want to make such a hugely game altering change at this late date?
[Edit: To show another example, my calculations show that the odds for a veteran Immortal or veteran Knight vs a veteran Spearman, unfortified on grassland, will change from 79.2%/20.8% to 97.2%/2.8%.
Game-altering indeed. Persia will be unstoppable
]@derekroth: I think you are on a right track to working out the new odds except for one complication which must be added: you need to work out and accumulate the odds for each separate sequence which results in an overall win or loss, i.e. must allow for W,W,L,W, W,L,W,W, etc. in the overall weighting of results. If you don't do that, the values for the W,W,W,W and L,L,L,L sequences influence the result out of the correct proportion.
And you also need to allow for the number of HP involved. Your calculation of 80%/20% for the tank is true only if both units have just 1HP. If they both have 3HP the result would change to 94.2%/5.8%.
robcheng
Sleep-deprived Addict
Quick question: "attacker wins" means that the attacker scores a 'hit' and takes off one of the defender's HPs correct? Are you taking this into account in your calculations? It seems like to be thorough we would need to create a table of possible outcomes (veteran attacker wins with all 4 HPs remaining: x%, with 3 HPs remaining: y%, etc.). This will probably mean you need to have significantly more defenders than currently, since you can't count on 1 or 2 spears to kill an attacking swordsman, you now may need 5 or 6...
Dragonlord
Fantasy Warlord
Great news! I add my thanks to Firaxis to all of the above!
I think the new RNG is good news! As I see it, it will reduce (not eliminate) the random factor in battles and thus emphasize strategy vs. the 'mass army' approach - which is a VERY good thing, IMO.
I've never liked having to send in huge stacks of units to make sure of victory (the WWI approach) and I've always absolutely hated the 'Tank vs. Spearman' losses. In fact, I've usually reloaded after extreme cases (yeah, yeah, I know.. no need to throw any rotten eggs!
)
NOW I may actually start playing Ironman games...
I think the new RNG is good news! As I see it, it will reduce (not eliminate) the random factor in battles and thus emphasize strategy vs. the 'mass army' approach - which is a VERY good thing, IMO.
I've never liked having to send in huge stacks of units to make sure of victory (the WWI approach) and I've always absolutely hated the 'Tank vs. Spearman' losses. In fact, I've usually reloaded after extreme cases (yeah, yeah, I know.. no need to throw any rotten eggs!
)NOW I may actually start playing Ironman games...
socralynnek
Civ & Hattrick addict
Your math approach is not correct.
You have to count every possible outcome, so don't forget the tie.
And you have to count every permutationof WWWL, i.e. LWWW is something different than WWWL.
In your example (Attacker winning 60%), you get in one round a chance of 47.52 % of winning (WWWW, WWWL, LWWW,WLWW,WWLW and add your results above) and a 34.56 % chance of a tie.
So your chance of winning the fight for 1 HP, you get 47.52 % + 34.56%*47.52%(for winning after 1 re-roll)+34.56%*34.56%*47.52%+... which is about 72.5 %.
(Assumed that a re-roll means that you do a turn of 4 results again, or is it more like a best of 5?, which is not clear from Tavis' post)
But in the end you are right, the rolls are now better for the one who is stronger!
Great news about the patch, but can someone now please explain how corruption works with the FP?
Does it work exactly like our original palace now. So when FP is built we will have two cities with rank 1 (Capitol and city with FP) and two cities with rank 2 etc.? Is RCP still penalised?
Does it work exactly like our original palace now. So when FP is built we will have two cities with rank 1 (Capitol and city with FP) and two cities with rank 2 etc.? Is RCP still penalised?
Does this apply to PTW too. I always thought the results occurred in non random runs, and it is intriguing to see this apparently confirmed. My own efforts playing at "random" sequence generation have also been plagued by non random sequences butting in. I don't suppose the detail of how combat is calculated is available.
TheNiceOne
Emperor
I'm really, really worried about the combat calculator change. As SirPleb excellently displayed, even small advantages as swordman vs spearman now becomes almost sure wins. This looks to me as a change that will hurt much more than it helps.
So please, please think twice here. Let me help you by suggesting the sensible solution :
Have a setting in the editor that says how many calculations that should be made for each combat round, and let the user set any number from 1 (as in current versions of the game) to 4 (as you suggest) or even 10.
1, which it has been all the time should be default.
So please, please think twice here. Let me help you by suggesting the sensible solution
:Have a setting in the editor that says how many calculations that should be made for each combat round, and let the user set any number from 1 (as in current versions of the game) to 4 (as you suggest) or even 10.
1, which it has been all the time should be default.
I think this is a bad idea and I hope it does not make it into the final patch. This overly complicates combat so there is no reasonable, convienient way for the player to calculate the odds of a victory or loss in a given situation. Not to mention, another version of my calculator... I'll scrap it after this change, I just can't keep up with them
Also, about "streaks"... They are in your mind. The combat is random and this has been confirmed. I believe this is an example of Atari and others catering to the irrational desires of the statistically uninitiated. Please, don't make this change...
IIRC, the change to the combat system from Civ II to Civ III (i.e.- The elimination of firepower, armor, etc.) was to create a more balanced game. No longer can you get a mass of tanks (armor in Civ II) and dominate the world. I think this change is unbalancing and is counter-productive to the stated goals of the Civ III designers, who wanted Civ III to be more balanced game than its predecessor (e.g.- the inclusion of resources, Scientific leaders, peaceful ways to win, etc.). It is a step backward in my eyes...
I will not patch the game if this change is made...
This does significantly change the game -- I would say that it is no longer safe to use a lone spearman for defense at the beginning of the game on any city. Won't be terrible -- just different. Coud limit expansion to some degree by human players, but allow the humans to more succesful wage war.
As to the game being unplayable -- there are thousands of people playing the game right now. That seems to be a misinformed generalization.
As to the game being unplayable -- there are thousands of people playing the game right now. That seems to be a misinformed generalization.
I also don't like the new combat rules at all. It effectively means that units which are weaker will have a MUCH harder time getting in even a single blow.
A far better solution which would preserve that likelyness of weaker units at least damaging stronger ones, while still lowering the chances for freak upsets would simply be to multiply all unit hit points by 4. Of course this would require that the rate of fire on all artillery units, as well as all heal rates be multiplied by 4 as well in order to keep everything balanced.
In any case, PLEASE make this new combat optional in the patch. Otherwise, I think I'd be better off staying with the corruption bugs.
A far better solution which would preserve that likelyness of weaker units at least damaging stronger ones, while still lowering the chances for freak upsets would simply be to multiply all unit hit points by 4. Of course this would require that the rate of fire on all artillery units, as well as all heal rates be multiplied by 4 as well in order to keep everything balanced.
In any case, PLEASE make this new combat optional in the patch. Otherwise, I think I'd be better off staying with the corruption bugs.
Nad
Known Troublemaker
Like others have said, I am EXTREMELY concerned with the combat results change.
Just from the examples displayed here and a few quick tests I've tried myself, this just seems abysmally wrong.
PLEASE Firaxis: I think you are repeating the mistake you made with corruption, implementing a "fix" that is ten times worse than the problem!!
If you are reading this, I would urge you strongly to drop this change until it can be tested properly...the rest of your patch sounds wonderful.
As you can no doubt see, there are plenty of Civ 3 players that could test the combat results. Please leave it out for now, if it is tested to work okay then you can add it in at a later date in one of the later patches.
What I am most concerned about is that this patch could make the game even more unplayable...if you mess up the combat this game will be useless!
Just from the examples displayed here and a few quick tests I've tried myself, this just seems abysmally wrong.
PLEASE Firaxis: I think you are repeating the mistake you made with corruption, implementing a "fix" that is ten times worse than the problem!!
If you are reading this, I would urge you strongly to drop this change until it can be tested properly...the rest of your patch sounds wonderful.
As you can no doubt see, there are plenty of Civ 3 players that could test the combat results. Please leave it out for now, if it is tested to work okay then you can add it in at a later date in one of the later patches.
What I am most concerned about is that this patch could make the game even more unplayable...if you mess up the combat this game will be useless!
Similar threads
