Is Barrage broken?
- Thread starter OTAKUjbski
- Start date
50 cats? You are joking I hope.Even I find that fairly astonishing. Assuming a 50/50 split between siege and stack defenders, that's 50 units at least, assuming all Cats have Accuracy. That's nothing in the Industrial Era when you can Draft, but that early?
I frequently find that the most effective solution against castles is to tech Steel and blow it away with Cannon. With Accuracy Cannon, you can even blow away the defenses and take the city in the same turn.
I frequently find that the most effective solution against castles is to tech Steel and blow it away with Cannon. With Accuracy Cannon, you can even blow away the defenses and take the city in the same turn.
Kind ofMore realistically, dropping a wall in 1-2 turns is not at all uncommon.
Yes, walls are a more reasonable proposition. The 50% defence takes 13 normal catapult bombardments (at 4% each). Take a typical stack (for me
) of 6 catapults. Round 1; 24% defence removed. Round 2; 24 defence removed and 4 left. Round 3; some catapults used (sacraficed 
) for collateral and the other units attack and take city.But if we had promoted 2 of those 6 catapults with Accuracy things happen a lot faster. Round 1; 32% defence removed and 18 left. Round 2; Use 2 catapults with Accuracy and one other to reduce defence to zero, use some of remaining catapults (sacraficed
) for collateral and take city with other units. As long as you have enough units left over for the assault a few Accuracy catapults can make warring quicker and more economical.
OTAKUjbski
TK421
re: Accuracy: This thread is about Barrage -- more specifically Barrage vs City Raider & Combat. I'm sorry, but I don't see the correlation between Barrage vs Accuracy, since they have completely different applications (one for damaging units and one for reducing city defenses) -- nor does the Accuracy discussion seem to traveling in any direction related to the Barrage promotion line. Please take the Accuracy discussion to a new thread and/or limit yourself to discussing Barrage in this thread. Many thanks ...
----
Oh ok ... I'm sorry. I guess my point is this:
With Barrage III, those two Trebs have a 51.69% chance of both dying. (6.79% both survive & 41.52% one survives)
With CR3, those two Trebs have only a 5.39% chance of both dying. (58.79% both survive & 35.82% one survives)
So my question is this, then: for an increased Maceman survival rate of 15.2% chance, is it worth sacrificing twice as many Siege Units?
When considering production costs, supply lines and war weariness, I don't think it's worth it.
That's exactly my point. The +% tooltips are highly misleading -- even incorrect.
There is a +whatever% increase in the collateral damage formula. However, it is to baseCombatStr, which is buried deep in the beginning of the formula. Thus, it does not equate to the expected +whatever% increase in iCollateralDamage -- the amount of damage in HPs!
The increase in [actual HP] collateral damage applied is more closely +10%, +20% & +40%!
Unfortunately, I wasn't privy to that "long time" knowledge ... and thus the reason I started this thread.
Useless promotions should be fixed. (or at least have their tooltips reflect the correct information)
It isn't a bug ... it's because of where the +whatever% is added that yields less-than-expected results. (see above)
Reference: CvUnit::collateralCombat
----
Oh ok ... I'm sorry. I guess my point is this:
With Barrage III, those two Trebs have a 51.69% chance of both dying. (6.79% both survive & 41.52% one survives)
With CR3, those two Trebs have only a 5.39% chance of both dying. (58.79% both survive & 35.82% one survives)
So my question is this, then: for an increased Maceman survival rate of 15.2% chance, is it worth sacrificing twice as many Siege Units?
When considering production costs, supply lines and war weariness, I don't think it's worth it.
That's exactly my point. The +% tooltips are highly misleading -- even incorrect.
There is a +whatever% increase in the collateral damage formula. However, it is to baseCombatStr, which is buried deep in the beginning of the formula. Thus, it does not equate to the expected +whatever% increase in iCollateralDamage -- the amount of damage in HPs!
The increase in [actual HP] collateral damage applied is more closely +10%, +20% & +40%!
Unfortunately, I wasn't privy to that "long time" knowledge ... and thus the reason I started this thread.
Useless promotions should be fixed. (or at least have their tooltips reflect the correct information)
It isn't a bug ... it's because of where the +whatever% is added that yields less-than-expected results. (see above)
Reference: CvUnit::collateralCombat
I dont understand why you wouldnt just bring enough siege to reduce the defenders to the minimum they can be damaged without needing to use barrage to get them there. Siege is what wins wars, everything else is just there to finish the nearly dead enemies off.
OTAKUjbski
TK421
I doubt you're you saying to send unpromoted siege into combat as lambs to the slaughter, so that should bring it back around full circle to whether those siege engines get promoted to CR or Barrage (even if only once to unlock the Accuracy promotion).
So far, I can find NO situation in which I'd want to use Barrage over City Raider.
Plus, the tooltip is misleading, because the amount of increased collateral damage is not +20%, +50% and +100% ... it's really more like +10%, +20%, +40%!!!
That's what this thread is about ... Barrage vs City Raider.
No need to get
with us The only reason the Accuracy discussion got started was that someone mentioned Barrage 1 is a pre-requisite for it. That was supposed to be a point in favour of Barrage, since Accuracy is at least useful, occassionally, in some situations, even if Barrage is a waste. Unfortunately, as I pointed out, it's not the only pre-requisite for Accuracy as CR is also one and that as we've seen is usually a much better promotion for any seige that is going to attack a city. And it is interesting that both CR and Accuracy are only useful against cities whereas Barrage is not. So we could conclude that a CR1 Acc catapult makes a lot more sense than a Bar Acc one.
End of Accuracy discussion.
The thread is certainly worthwhile but barrage has never shown any worthwhile results in any of my games or any games I've seen published. I'm a little surprised that you didn't notice the poor performance in casual play, it was certainly apparent to me even though I never carried out an exhaustive study. I notice that the AI uses the promotion a lot for its seige; but since when has that been any recommendation? Perhaps that's what it's for, to weaken the AI's seige by giving it a useless promotionUnfortunately, I wasn't privy to that "long time" knowledge ... and thus the reason I started this thread.
@Otakujbski. I think accuracy was introduced to the discussion because the poster thought barrage was a needed promotion for it. And you know how this forum loves a tangent. It is an immutable law of nature and these forums that all threads will invariably end up as arguments of SE vs CE.
I shall now demonstrate.
With an SE you would be able to run vassalage for the extra exp to get accuracy because you do not typically use Bureaucracy. And the troop cost of Pacifism is offset partially by the free units from vassalage.
To which the reply is: But in a CE, you run Bureaucracy with Theo, The troop cost for Pacifism greatly outweighs any savings you get with Vassalage. And under Bureaucracy you generate more cash to support the war.....Let the argument begin.
I shall now demonstrate.
With an SE you would be able to run vassalage for the extra exp to get accuracy because you do not typically use Bureaucracy. And the troop cost of Pacifism is offset partially by the free units from vassalage.
To which the reply is: But in a CE, you run Bureaucracy with Theo, The troop cost for Pacifism greatly outweighs any savings you get with Vassalage. And under Bureaucracy you generate more cash to support the war.....Let the argument begin.
OTAKUjbski
TK421
@ CivCorpse: And so is affected the price of tea in China ...
actually now that there is no longer a Chinese leader in civ, the price of tea is irrelavent
OTAKUjbski
TK421
Barrage vs CR (Trial 1: Cats & Swords vs Longbows)
Okay ... so I ran my first battery of trials today and have what I interpret as moderately inconclusive results.
The first thing I noticed was because of the RNG's fickleness, I need to narrow down the scope of my test parameters (eliminate some variables) and run the test many, many more times (at least 100 per promotion combination would be optimal, but I'm not certain I have the stamina for that).
Similarly, I scrapped my first 200 test results after seeing an unexpected symmetry across them. I narrowed it down to the lack of a single game option: New Random Seed on Reload. I was reloading the same testbed game between trials, which resulted in a lack of "random" in the number generator.
All 'real' tests have the New Random Seed on Reload option turned ON.
----
In each test, a stack of Siege units attacks a city defended by non-Siege units on flat land with 0% city defenses. Afterwards, a stack of non-Siege units attacks the same city. The number of surviving attack units is recorded.
This test is run with 3 promotion variables, resulting in 140 total combinations of attacking and defending strengths:
I've recorded the results into an Excel spreadsheet, which is initially pretty daunting to gaze upon, so I'll try to explain it briefly here:
----
So onto the tests ...
6x Catapult, 12x Swordsman vs 6x Longbowmen
3x Catapult, 12x Swordsman vs 6x Longbowmen
@ ALL: QUESTIONS, DIRECTION NEEDED
These tests are yet inconclusive -- particularly because only 3 iterations isn't enough to establish a reliable pattern. (Future tests will be run at least 10 times each.)
I will be re-running this set of tests as well as (hopefully) a couple more tomorrow, but I need to know which variables to use.
Anything else I might've missed @ 0130 ...
BTW, I'll run Tank tests later.
Okay ... so I ran my first battery of trials today and have what I interpret as moderately inconclusive results.
The first thing I noticed was because of the RNG's fickleness, I need to narrow down the scope of my test parameters (eliminate some variables) and run the test many, many more times (at least 100 per promotion combination would be optimal, but I'm not certain I have the stamina for that).
Similarly, I scrapped my first 200 test results after seeing an unexpected symmetry across them. I narrowed it down to the lack of a single game option: New Random Seed on Reload. I was reloading the same testbed game between trials, which resulted in a lack of "random" in the number generator.
All 'real' tests have the New Random Seed on Reload option turned ON.
----
In each test, a stack of Siege units attacks a city defended by non-Siege units on flat land with 0% city defenses. Afterwards, a stack of non-Siege units attacks the same city. The number of surviving attack units is recorded.
This test is run with 3 promotion variables, resulting in 140 total combinations of attacking and defending strengths:
- The attacking Siege units have: (1) no promotion; (2) Barrage I; (3) Barrage I, II; (4) Barrage I, II, III; (5) City Raider I; (6) City Raider I, II; or (7) City Raider I, II, III.
- The attacking non-Siege units have: (1) no promotion; (2) City Raider I; (3) City Raider I, II; or (4) City Raider I, II, III.
- The defending non-Siege units have: (1) no promotion; (2) City Garrison I; (3) City Garrison I, II; (4) City Garrison I, II, III; or (5) Drill I, City Garrison I, II, III.
I've recorded the results into an Excel spreadsheet, which is initially pretty daunting to gaze upon, so I'll try to explain it briefly here:
Spoiler Excel spreadsheet legend :
----
So onto the tests ...
6x Catapult, 12x Swordsman vs 6x Longbowmen
Given the parameters above, each combination was tested 3 times (which isn't nearly enough) with the following results:
However, this test is innately biased against Barrage.!? Since the number of attacking Siege units is equal to the number of defenders, each defending unit is most likely to enter direct combat with a Siege unit ...
... which led to my next test (which will hopefully bias in favour of Barrage) ...
Spoiler results for 6x Catapult, 12x Swordsman vs 6x Longbowmen :
- In every case where the # of Siege and non-Siege promotions was equal, promoting down the City Raider line resulted in similar or better results -- particularly in the survival rate of the attacking Siege units.
- When the attacking non-Siege start with good or better Combat Odds (i.e. weak defenders), promoting Siege units down the City Raider line results in a better overall survival rate -- particularly to the attacking Siege units.
- When the attacking non-Siege start with poor or worse Combat Odds (i.e. strong defenders), promoting Siege units down the Barrage line shows promise -- even if less than expected.
However, this test is innately biased against Barrage.!? Since the number of attacking Siege units is equal to the number of defenders, each defending unit is most likely to enter direct combat with a Siege unit ...
... which led to my next test (which will hopefully bias in favour of Barrage) ...
3x Catapult, 12x Swordsman vs 6x Longbowmen
Spoiler :
- When the number of attacking Siege units is less than the number of defenders, Barrage continues to show promise, but still not as much as I was hoping for. (In most cases where the number of Siege and non-Siege promotions were equal, the comparison is inconclusive, IMO.)
- Promoting attacking Siege units down the City Raider line considerably increases their individual chances of survival.
- Promoting attacking Siege units down the Barrage line appears to increase the attacking non-Siege units' chances of survival.
- When the attacking non-Siege units begin with a snowball's chance of surviving, Barrage appears to be the better choice.
@ ALL: QUESTIONS, DIRECTION NEEDED
These tests are yet inconclusive -- particularly because only 3 iterations isn't enough to establish a reliable pattern. (Future tests will be run at least 10 times each.)
I will be re-running this set of tests as well as (hopefully) a couple more tomorrow, but I need to know which variables to use.
- Which attacking Siege & non-Siege units should I use against which defending non-Siege units? (I want 3 sets)
I'm thinking: (1) Catapult, Swordsman vs Longbowman; (2) Cannon, Rifleman vs Rifleman; (3) Artillery, Infantry vs Infantry. - Which defending promotions should I use? (I want 3 sets)
I'm thinking: (1) CG1; (2) Drill 1, CG1; (3) Drill 1, CG1, CG2, CG3 -- which roughly represents non-Protective, Protective and Sitting Bull. - What quantity of units should I use?
I'm thinking: (1) 6 Siege, 12 non-Siege vs 6 non-Siege defenders; (2) 3 Siege, 12 non-Siege vs 6 non-Siege defenders
Anything else I might've missed @ 0130 ...
BTW, I'll run Tank tests later.
siggboy
Monarch/Epic
Otaku, I'm impressed by the amount of zeal that you're currently putting into your experimenting, but you'd have to draw a lot of independent samples to come to definite conclusions. Given the number of combinations for troop promotions even if you keep it down to just the reasonable scenarios this is quite infeasible.
I will write a complete combat simulator for Civ because I'm also interested in some definite answers to the question at hand and others (regarding the value of "Drill", for example). With the simulator it will be possible to do thousands of test battles between two given units stacks, hopefully approaching an exact model close enough.
I hope I'll get to a useable version during the weekend, I will post here when I've got something.
As far as Barrage vs. City Raider for siege weaponry goes: one aspect that has not been mentioned yet are Great General Points and earned XP. Considering the far higher survival rate for CR siege, this can add up to a sizeable amount of GPP over the course of a game.
Siege survival rate directly translates into GPP: if you go on campaign with x siege weapons, and 50% of them survive, you've earned x/2 GPP from these fights, compared to x/5 at a 20% survival rate (I'm using the numbers from B2 and CR2 trebs vs. D1/CG3 Longbows here). I'm aware that it's a simplified scenario for many reasons.
The CR2 trebs earn you 2.5 times as many GPP as the B2 trebs (and your opponent get 2.5 times less GPP because your winning fights are his losing fights).
Given that siege weaponry is making up a significant portion of a typical army this translates into a lot of GPP during a given war. The amount of units saved from additional Barrage collateral damage would have to be significant to make up for that.
Earned XP can also be valuable, especially at the point where you can make, say, 9 XP siege weaponry and just one winning fight will make them immensely stronger due to the additional promotion. This is even better for charismatic leaders.
I will write a complete combat simulator for Civ because I'm also interested in some definite answers to the question at hand and others (regarding the value of "Drill", for example). With the simulator it will be possible to do thousands of test battles between two given units stacks, hopefully approaching an exact model close enough.
I hope I'll get to a useable version during the weekend, I will post here when I've got something.
As far as Barrage vs. City Raider for siege weaponry goes: one aspect that has not been mentioned yet are Great General Points and earned XP. Considering the far higher survival rate for CR siege, this can add up to a sizeable amount of GPP over the course of a game.
Siege survival rate directly translates into GPP: if you go on campaign with x siege weapons, and 50% of them survive, you've earned x/2 GPP from these fights, compared to x/5 at a 20% survival rate (I'm using the numbers from B2 and CR2 trebs vs. D1/CG3 Longbows here). I'm aware that it's a simplified scenario for many reasons.
The CR2 trebs earn you 2.5 times as many GPP as the B2 trebs (and your opponent get 2.5 times less GPP because your winning fights are his losing fights).
Given that siege weaponry is making up a significant portion of a typical army this translates into a lot of GPP during a given war. The amount of units saved from additional Barrage collateral damage would have to be significant to make up for that.
Earned XP can also be valuable, especially at the point where you can make, say, 9 XP siege weaponry and just one winning fight will make them immensely stronger due to the additional promotion. This is even better for charismatic leaders.
TheMeInTeam
If A implies B...
- Joined
- Jan 26, 2008
- Messages
- 27,989
If you're inclined to run more tests, I'd be interested to see how barrage performs when fighting huge stacks, say 30-40 or more. If your goal is to wipe out those enemy units with as few losses as possible, and it's truly a compressed enemy stack (I've seen montezuma/shaka field stacks like this), barrage might be more appealing. The scenario sounds ridiculous but it does occur. This is especially worth considering if the siege unit is fighting in the field with already equal/greater strength, such as the cannons vs knights scenario (or even cannon vs rifle, where combat II isn't going to put the cannon over the top vs any kind of promo on the rifle). I think such a scenario would also be a worthy consideration when considering what promotions to give tanks.
Really, that's the only time I can envision the small bonuses provided by barrage adding up enough. If there are THAT many units, it seems like it'd have use. Casual play suggests that to be the case to me, but going by feel vs objective proof can be misleading.
Really, that's the only time I can envision the small bonuses provided by barrage adding up enough. If there are THAT many units, it seems like it'd have use. Casual play suggests that to be the case to me, but going by feel vs objective proof can be misleading.
obsolete
Deity
The only thing broken in Barrage is that silly +10% vs meele. Now assuming you can get promoted all the way up to Barrage III, you are out of the meele ages anyhow.
OTAKUjbski
TK421
re: Barrage isn't broken
So even though the Barrage I, II & III tooltips claim "+20% ...", "+30% ..." and "+50% collateral damage" but really only increase collateral damage by a maximum (it fluctuates) of +12.5%, +12.5% and +25% ... that's not broken?
Or perhaps how in some cases with all Siege Weapons, the difference between the Barrage n and Barrage n+1 promotions actually equates to +0% ... that's not broken, either?
How do you figure either of those cases isn't broken?
re: combat simulator
Me too ... I can't believe I'm doing this instead of playing.
As long as we're willing to make inferences based on emerging patterns, I think these 'limited' results can be conclusive ...
... but those would be considerably more conclusive!!!
I greatly look forward to seeing it.
re: GG GPP
Well, in fairness to Barrage, I think the disparity between GG GPP is because Barrage is broken. After all, if Barrage worked as it seems to have been intended, the lower survival rate of the Siege weapons would come in exchange for a similarly higher survival rate of the non-Siege weapons.
But as it is now, Barrage appears to do little/nothing by comparison to the City Raider promotion. (Maybe it'll fair better compared to Combat versus troops in the open field.?)
re: bigger stacks
Gimme the test parameters, and I'll run them.
efender ratio was 1:2 ... the ratio would likely have to be an unrealistically low 1:6 ratio before any real divergence appears.
So even though the Barrage I, II & III tooltips claim "+20% ...", "+30% ..." and "+50% collateral damage" but really only increase collateral damage by a maximum (it fluctuates) of +12.5%, +12.5% and +25% ... that's not broken?
Or perhaps how in some cases with all Siege Weapons, the difference between the Barrage n and Barrage n+1 promotions actually equates to +0% ... that's not broken, either?
How do you figure either of those cases isn't broken?
re: combat simulator
Me too ... I can't believe I'm doing this instead of playing.
As long as we're willing to make inferences based on emerging patterns, I think these 'limited' results can be conclusive ...
... but those would be considerably more conclusive!!!
I greatly look forward to seeing it.
re: GG GPP
Well, in fairness to Barrage, I think the disparity between GG GPP is because Barrage is broken. After all, if Barrage worked as it seems to have been intended, the lower survival rate of the Siege weapons would come in exchange for a similarly higher survival rate of the non-Siege weapons.
But as it is now, Barrage appears to do little/nothing by comparison to the City Raider promotion. (Maybe it'll fair better compared to Combat versus troops in the open field.?)
re: bigger stacks
Gimme the test parameters, and I'll run them.
- What quantity and type of attacking Siege units?
- What quantity and type of attacking non-Siege units?
- What quantity and type of defending units?
- What battlefield terrain conditions?
efender ratio was 1:2 ... the ratio would likely have to be an unrealistically low 1:6 ratio before any real divergence appears.BarrageII provides the bonus vs melee. You can get barrageII from a barracks and 1 civic. BarrageIII provides a 10% bonus vs gunpowder units.
Barrage is best for dealing with an AI SOD
CR is best for city promotions.
2 other factors against barrage when you attack a city:
1. Giving up CR even on a suicide unit decreases the damage done to the top defender. This is hard to calculate but I would argue of some significance.
2. When the AI has a big stack, it's normally a mixed stack. So a 10 unit AI stack that you attack with cats might be 5LB, 1pike, 2 cat, 2 HA--something like that. Presumably you have mace. Only the LB really are a threat, and the cat are immune to collat. So any collat that hits a seige unit is wasted, and increasing collat on weak units is mostly wasted as its overkill.
1. Giving up CR even on a suicide unit decreases the damage done to the top defender. This is hard to calculate but I would argue of some significance.
2. When the AI has a big stack, it's normally a mixed stack. So a 10 unit AI stack that you attack with cats might be 5LB, 1pike, 2 cat, 2 HA--something like that. Presumably you have mace. Only the LB really are a threat, and the cat are immune to collat. So any collat that hits a seige unit is wasted, and increasing collat on weak units is mostly wasted as its overkill.
vicawoo
Chieftain
- Joined
- Feb 12, 2007
- Messages
- 3,226
Lot of information to process. Well, with 6 catapults, you're averaging less than 1 swordsman death (probably the 11's are within the standard deviation of losses). Which means: if the swordsmen deaths are almost constant, then the only that matters are your catapult losses.
Your catapult losses numbers are too small in sample size, you could use some relative risk type cross comparisons to get more consistent data. And the collateral damage cap is 50%, so with barrage 1, you could use 4 catapults and still do almost half damage. Yeah yeah, primary defender exceptions, but you should automatically have 2 less catapult deaths.
Your catapult losses numbers are too small in sample size, you could use some relative risk type cross comparisons to get more consistent data. And the collateral damage cap is 50%, so with barrage 1, you could use 4 catapults and still do almost half damage. Yeah yeah, primary defender exceptions, but you should automatically have 2 less catapult deaths.
OTAKUjbski
TK421
Forgive me if I misunderstand these ...
I think this is [only] true when the non-Siege attackers begin with a good chance of survival.
In the Lvl1 Swords vs D1/CG3 Longbows, even after 6 Lvl1 Cats, the loss rate averaged -25%.
In BtS, collateral damage is capped at 75%, 80% and 85% (Cats, Cannons, Artillery) since Siege units can't kill anymore.
Are you suggesting I increase the quantity of all units? So maybe 12 Cats & 48 Swords vs 24 Longbows?
I think this is [only] true when the non-Siege attackers begin with a good chance of survival.
In the Lvl1 Swords vs D1/CG3 Longbows, even after 6 Lvl1 Cats, the loss rate averaged -25%.
As long as I'm protecting my heavy hitters, 25% losses are acceptable, in which case Barrage vs City Raider does come down to the survival rate of the Siege.
But in the 2nd example, after sending in just 3 Lvl1 Cats against D1/CG3 Longbows, the loss rate of the Lvl1 Swords went up to 75% -- in which case the goal is either minimizing non-Siege losses (presumably via Barrage) or minimizing Siege losses or balancing all losses (presumably via CR).In BtS, collateral damage is capped at 75%, 80% and 85% (Cats, Cannons, Artillery) since Siege units can't kill anymore.
Are you suggesting I increase the quantity of all units? So maybe 12 Cats & 48 Swords vs 24 Longbows?
