Elite Troops die easier
- Thread starter Ozymandius
- Start date
TheNiceOne
Emperor
I don't doubt that there are numerous situations where an army performs worse than expected, but I'll bet that there are just as many situations where armies perform better than expected.
You first post that it is obvious that armies lose HP more frequently, and then tell that you selmdom have armies, i.e. your statistical base is very small.
Since this is a statement that can be verified without any problem (save before army attack, then reload and attack with another unit), but has not been verfied by anyone, I seriously doubt that you have seen anything else than normal randomness.
TheNiceOne
Emperor
**** happens.
In my current game I've lost a 12 HP cavalry army to a 4 HP rifleman in a 7+ size city, I think the rifleman only lost one HP. But I also took out another rifleman with another army without losing a single HP. Another army attcked a rifleman and lost 11 HP and only took 1 HP from the rifleman, then took out the rifleman's remaining 3 HP.
The only thing this proves is that when a game is random, sometimes you're lucky and sometimes you're not.
Masquerouge
Deity
AMEN, my brother, thou spoke well !
I would add verse 2 : "1. For it shall be so, that in random games, luck shall come to you. 2. 50% of the times, no more, no less, it shall come to you. 3. 40% it will come to you only if 60% it comes to you after. 4. Fiat Alea ! Said the Lord. 5. And he saw that it was fair."
Ozymandius
Prince
Hey there nice one
The next time my elite unit loses when a regular one does not I'll save the file for you.
If you think that the AI doesn't cheat..
Trade away communicatioins or advance with a civ and find that other civ's have it before you can contact them while still on your turn.. don't have the chance to sell it and profit myself but even the pooorest civ's have it
If you have a choke point that you can close off using troops, the AI knows about even when ten or so tiles away. Just try it. If the AI likes to cross your territory and you can block, it will turn around, open it and it wil turn again.
Kill another civ early and it spawns stronger than before, workers transport to the new location from their previous city and they have a second settler with a spearman rather than warriors.
The next time my elite unit loses when a regular one does not I'll save the file for you.
If you think that the AI doesn't cheat..
Trade away communicatioins or advance with a civ and find that other civ's have it before you can contact them while still on your turn.. don't have the chance to sell it and profit myself but even the pooorest civ's have it
If you have a choke point that you can close off using troops, the AI knows about even when ten or so tiles away. Just try it. If the AI likes to cross your territory and you can block, it will turn around, open it and it wil turn again.
Kill another civ early and it spawns stronger than before, workers transport to the new location from their previous city and they have a second settler with a spearman rather than warriors.
Random Passerby
Bystander
- Joined
- Jan 3, 2002
- Messages
- 95
If you define "AI cheating" as the AI doing things and knowing things that the human player cannot, then I would hate to think how huge the list of "human cheats" would be.
punkbass2000
Des An artiste
This would prove nothing. If you repeated this (without 'random seed preserved' on) 100 times and the regular won more often than the elite, I would consider believing you. Otherwise you're just demonstrating the all too common trait of selective memory with your theory.
If you think that the AI doesn't cheat..
Trade away communicatioins or advance with a civ and find that other civ's have it before you can contact them while still on your turn.. don't have the chance to sell it and profit myself but even the pooorest civ's have it
If you have a choke point that you can close off using troops, the AI knows about even when ten or so tiles away. Just try it. If the AI likes to cross your territory and you can block, it will turn around, open it and it wil turn again.
Kill another civ early and it spawns stronger than before, workers transport to the new location from their previous city and they have a second settler with a spearman rather than warriors.
Everyone knows the AI cheats. There have been many threads on it. However, the fact that the AI cheats in these ways does not mean the AI plays with the random number, and many people have proved that by doing one thousand combats and such to prove the 'spearmen beat my tank often' crowd.
TheDS
Regular Riot
Hey hey hey, people!
Everyone knows that whenever the odds of a thing are 50/50, the undesireable thing will happen 90% of the time. It's a Murphy's Law, and those are involiable.
While I haven't gone to the extreme lengths of spending half a day to reload a game and mark the results, and then spend the rest of the month doing this over and over, I have noticed that with startling regularity I will lose battles that are heavily in my favor.
In the real world, a 6-to-1 battle is virtually a guarantee of victory. This doesn't mean 6 guys to 1, but when you take into account things like training, equipment, morale, and most important of all, surprise, if you have 6-to-1 odds or better, you win.
But I've seen a number of battles in one game where I had 8-to-1 or even better odds, and still lost! No, this isn't every single combat. Maybe it's 1 in 100, or 1 in a 1000. But when I have an ATK of 8 and the enemy has a DEF of 1 and he's on open ground and not fortified and I have twice as many hit points, I should NEVER see a loss. NEVER. Well, maybe once per game, but that's it.
Instead, I see it several times a game. For me, I have to chalk it up to my personal bad luck. I can't roll dice for $hit either, so I often seize defeat from the jaws of victory whilest playing Risk or A&A.
AAAAAAAnyway..... Now you know what's the deal. It's your personal bad luck, not the game's fault for picking up on it.
Everyone knows that whenever the odds of a thing are 50/50, the undesireable thing will happen 90% of the time. It's a Murphy's Law, and those are involiable.
While I haven't gone to the extreme lengths of spending half a day to reload a game and mark the results, and then spend the rest of the month doing this over and over, I have noticed that with startling regularity I will lose battles that are heavily in my favor.
In the real world, a 6-to-1 battle is virtually a guarantee of victory. This doesn't mean 6 guys to 1, but when you take into account things like training, equipment, morale, and most important of all, surprise, if you have 6-to-1 odds or better, you win.
But I've seen a number of battles in one game where I had 8-to-1 or even better odds, and still lost! No, this isn't every single combat. Maybe it's 1 in 100, or 1 in a 1000. But when I have an ATK of 8 and the enemy has a DEF of 1 and he's on open ground and not fortified and I have twice as many hit points, I should NEVER see a loss. NEVER. Well, maybe once per game, but that's it.
Instead, I see it several times a game. For me, I have to chalk it up to my personal bad luck. I can't roll dice for $hit either, so I often seize defeat from the jaws of victory whilest playing Risk or A&A.
AAAAAAAnyway..... Now you know what's the deal. It's your personal bad luck, not the game's fault for picking up on it.
TheNiceOne
Emperor
Hey to you
Do that, I'll be very interested in seeing it. But remember that "save random seed" must be on. If not, it proves absolutely nothing.
I haven't seen this in the latest two patches, so I would like to see a save game with this as well. If I contact other civs and sell them techs during my turn, I always get to sell it to everyone first.
I know. This is a cheat, but remember that "memory" is damned hard to program. If the AI didn't know the position of your units, it would need a memory of where and when it saw some of your units etc. Instead of undertaking this difficult job, Firaxis gave the AI knowledge of the position of all your units.
Not a cheat. This is a game rule, and after 1.21 you can toggle it on or off before you start a game.Kill another civ early and it spawns stronger than before, workers transport to the new location from their previous city and they have a second settler with a spearman rather than warriors.
I haven't said that the AI doesn't cheat, but I have said, and believe that the AI does not cheat during battle resolution. Again, check etj4Eagle's excellent thread: http://forums.civfanatics.com/showt...&threadid=21378
etj4Eagle
ACME Salesman
Ah Axis and Allies and Risk. I have seen die rolls in those games for more improbable than the spearman beating the tank resolutions. Of course I guess seeing how the dice fall make us more okay with the results, whereas here they are hidden.
And back to the original poster, there is nothing funky in the combat resolution code. It follows all the rules correctly without any special bonus to AI units. However it does use a random number generator, one that is perfectly fine, but remember that any good string of random number will be streaky.
Had two spearman defend against a stack of (I think) 9 units, including three swordsmen, a couple warriors and the rest were archers. Fully expected to lose the city and yet didn't lose either spearman.
I was so outraged by these results that I promptly reloaded and lost the city, the way it should have been originally.
Well, the first paragraph is true, anyway.
I was so outraged by these results that I promptly reloaded and lost the city, the way it should have been originally.
Well, the first paragraph is true, anyway.
Masquerouge
Deity
And I can distinctly remember (because it was last night) that during my full-scale invasion of France, in which I took 5 cities including Paris, and killed something close of 150 french units with my 60 cavalry units, I lost THREE (not one, not two, THREE) cavalry to spearmen. (I lost 10 cavalry on the whole).
If that is not AI cheating ?
Yeah, selective memory is such a bad thing.
I just wish there was a way to see the units you lost, the units you crushed and against whom, like in CTP...
If that is not AI cheating ?
Yeah, selective memory is such a bad thing.
I just wish there was a way to see the units you lost, the units you crushed and against whom, like in CTP...
Maple
Canadian Patriot
Think about it... To bypass the complicated calculations in battle results, it is surprisingly close to: If a 24attack unit attacks a 2defend unit, then the chances to defend succesfully are about 1/12. (2/24=1/12)
Cavalry have an attack of six. Spearmen defend at two. They have a 33% chance to defend against attacks. Masquerouge, just how many different spearmen did you attack? 20? 40? 80?
Cavalry have an attack of six. Spearmen defend at two. They have a 33% chance to defend against attacks. Masquerouge, just how many different spearmen did you attack? 20? 40? 80?
Ozymandius
Prince
my point is:
If you have a choice of attacking units and one of your units is elite... save first ...
attack with elite... who dies. Reload
attack with veteran who lives.
reload try again (Random? seed.)
attack with elite, who dies, reload, attack with other veteran who lives?
I think we are expected to attack with elite in the hope of getting a leader but the result is frequent losses of our finest..
If a 4 striper can beat a three, why can't a five striper?
If you have a choice of attacking units and one of your units is elite... save first ...
attack with elite... who dies. Reload
attack with veteran who lives.
reload try again (Random? seed.)
attack with elite, who dies, reload, attack with other veteran who lives?
I think we are expected to attack with elite in the hope of getting a leader but the result is frequent losses of our finest..
If a 4 striper can beat a three, why can't a five striper?
punkbass2000
Des An artiste
You will always get the same results if you don't turn off 'Preserve Random Seed'.
Zachriel
Kaiser
1 in 3.5 in Masquerouge's example.
Possible factors involved in the slight variation: number of hitpoints; the exact tactical situation, such as walls, fortified, river or city; or possible enemy counterattacks.
Nary a war has been fought without blood shed on both sides.
Zachriel
Kaiser
Elites are stronger. Masquerouge may be extrapolating from too little data. The reason we know is because we can experiment with the hitpoint values in the editor.
The mod community is well aware of the powerful influence of fiddling with the hit point values. As you raise the hitpoints, the more certain victory becomes for the superior force. So swordsmen win more often versus warriors, and tanks never lose to spearmen. Increasing the hitpoints also exaggerates the effect of terrain. In other words, the more the hitpoints the more closely the results resemble the expected.
Due to the way the combat resolution occurs, each hitpoint is resolved independently -- even if the randomizer is faulty.
Check out this civulator and try some high hitpoint values.
http://207.191.20.18/civ3/lwc-civulator.html
Then try to mod them in the editor. Warning! This procedure can unbalance the game in unexpected ways.
etj4Eagle
ACME Salesman
Assuming that you have not deactivated the saving of the random seed. You will not see the differance that you are claiming. Attack with an elite and it dies. Reload the game and repeat with a veteran instead and guess what it dies as well. In fact all the round resolutions up to the time that unit dies will be identical.
From my other tests I have seen this to be true. However, you have to make sure that you don't use up extra random numbers the second time by doing things not exactly the same.
The game only cares about round by round resolution, it does not care about what classification the unit is.
Now if you don't have that random seed preserved then this whole exercise goes out the window. Since instead of generating a series of die rolls that in a swordman-spearman match up (for ex) would give W-L-L-L-W-L-L. The second time you might get W-W-W. Now with random seed preserved when you do the roload you will notice your regular unit still dies, but only does 1pt of damage verse the 2 that the elite would have done.
Ozymandius
Prince
After loading from a save the same result is arrived at by doin the same action in the ames sequence.
Goodie huts for example:
enter the village from 5 saves and reloads givesthe same result .
wait a return and normally get a different result.
Elites do not always die, but with more regularity than randomness should allow ie. on the same throw a weak unit should not survive while a stronger unit dies.
I play only on Pangea with all options engaged. I do not edit the rules but I do use save/reload when David slays Goliath. I do not reload when the enemy units are superior to my own
Goodie huts for example:
enter the village from 5 saves and reloads givesthe same result .
wait a return and normally get a different result.
Elites do not always die, but with more regularity than randomness should allow ie. on the same throw a weak unit should not survive while a stronger unit dies.
I play only on Pangea with all options engaged. I do not edit the rules but I do use save/reload when David slays Goliath. I do not reload when the enemy units are superior to my own
TheNiceOne
Emperor
Too bad you haven't bothered to mark the results. Until you have some facts, I'll rather believe etj4Eagle who has proved that the battle resolution follows the statistically expected results.
Try etj4Eagle's combat resolution save game, and then post your results when you too have some facts. Honestly, you sound like some of the whiners I know from playing ASL, that start whining about their bad luck after the second dice roll of the game.
Similar threads
