Battle Mechanics

[Joshua84]

Alright, does anybody know how the current battle mechanics work or are supposed to work? From what I have seen and experienced, the game heavily favors an invading army. Most recently, my civilization was invaded. We started off about equal, they had maybe 20 more points than we did starting off. I switched to a fortify position on my troops for a round in order to prevent them from dealing too much damage. Then switched to Heroic stances during the first round of combat thinking there were not many troops killed so we should have a turn. Well, the opposing army got an extra attack. I figured it was OK since they invaded. My civilization got to attack after 3 rounds of them destroying my troops, when they did not outnumber us by much to start with. Well, later rounds I ordered a retreat since the opposing army was getting 12 and 13 attacks with no sign of letting up even though we still had over half the points left. Their point total was around 550 and ours was about 285.

Any ideas on how the battle mechanics work? I enjoy CivWorld, but don't know if it is worth it when battle mechanics seem so unbalanced.

[robin74]

They got more attacks because you went fortified, not because they were in an offensive war. Fortifying makes your chances to attack about 4 times lower compared to heroic.
You say they didn't outnumber you much. What do you mean by that? The only number that matters is your strength - the number in the world battle screen (and up to the rainy/windy bug, the one that should also be displayed on the left of your battle screen). When you go fortified, you cut this number by half (compared to normal). If your strength is lower than theirs, they are more likely to attack. If you fortify, they are much more likely to attack.

Their point total was around 550 and ours was about 285

I'd say you were strongly outnumbered.

[Joshua84]

I'd say you were strongly outnumbered.

When I had it at heroic, the point difference was about 8. The number on their side of the bar read 933.8 or so, whereas ours read 925.2 I think. The exact decimal is questionable, but the main numbers are correct. They had an attack, and dropped the number down to 915-ish. They had another attack, which dropped it to around 903. Then they received a third attack, which dropped it to the 890's. Then we managed to get 1 attack back. The game should have only allowed 1, maybe 2 attacks before we had an attack at that initial point difference. Only after they had 3+ attacks to our 1 did I switch to fortify, thinking that they could wittle our numbers away more slowly before we had a chance to attack.

[ShuShu62]

There are a number of unknown factors. You cannot assume a straight 50/50 chance if the sides are evenly matched. Interestingly, your suprise at getting hit 3 straight times indicates that your chance of getting hit both the second and third shot were considerably less than 50%.

Straight 50/50 probability would say that you would see an opponent take a second shot half the time after the first shot. Nobody see's that behavior. so the odds are not 50/50. There is a bias towards trading shots, but it is not absolute.

If it were really 50/50 the odds of you getting hit a third time after taking two hits would be 50/50. Why would you be shocked to get hit a third time after watching yourself get hit twice? Real answer is because the odds are so biased in favor of trading shots that losing 3 shots in a row in an even battle is rare.


---- There is the other side of the coin as well---
Beer Bet: We are in a bar and I flip the coin once and it comes up heads.

Should you bet on heads or tails for the next flip?

[robin74]

The game should have only allowed 1, maybe 2 attacks before we had an attack at that initial point

Why?
Who gets an attack is random. With your strength being roughly equal, you would get to attack roughly half of the time on average. Getting three heads in a row in a coin toss isn't rare or unlikely. You just had a little of bad luck. And you should have added more troops, especially once you took losses, not to mention you should never allow starting with strength 8 points lower - buying 10 more points should really be easy.

before we had a chance to attack.

Once you fortified your chances to get an attack decreased considerably.

[SJC]

I'm pretty sure that whoever has the higher number has a lot greater chance of getting the next attack. Especially early in the battle, if it is relatively close, the lesser army does have a significant chance of attacking, but the larger army will get most of the attacks.

As the differential grows, the smaller army will get fewer and few attacks. This is one of the reasons that getting as heroic as possible is a good idea (the other is that you hit for more damage). If you can get over 2-1 (no idea if this is a hard target), the lesser army will get a lot fewer attacks. Around 10-1, they will get almost none.

Fortified makes the differential a lot larger, highly increasing the chance that you will not get the next attack. Fortifying can be useful for placeholder units (such as attack 1 units that are there to keep a section of the field occupied). As well, it can be useful if you think waiting for a weather change will give you a chance. EDIT: another use is to cause another section of your army to attack -- fortify your infantry to make your horses attack, for example (more on that tactic in other threads).

If it is close, you need to do something to push the arrow in your direction. Run the tower game again or buy some units on the market. Make all your units heroic. If you have a civbuck, call up a storm (or whatever sort of weather you need). If you buy or build units, wait until just before the turn changes so they don't have time to respond - and hope they aren't doing the same.

Having hammers a the beginning of the battle that are kept in reserve can be a good idea. Once you see how big an army the enemy has brought to the field, and how it is set up between horses, melee, bows, etc. you can use those hammers most effectively.

Beer Bet: We are in a bar and I flip the coin once and it comes up heads.
Should you bet on heads or tails for the next flip?

We know nothing about the coin or where it came from. Perhaps its a trick coin or a perfectly legitimate one. It could be accidentally weighted due to poor production processes. However, the evidence we do have is that it CAN come up heads.

[ShuShu62]

...
We know nothing about the coin or where it came from. Perhaps its a trick coin or a perfectly legitimate one. It could be accidentally weighted due to poor production processes. However, the evidence we do have is that it CAN come up heads.

SJC is saying he would bet heads, and he would be correct. Civworld combat is the coin. Things we have seen...

1) The highest strength takes the first battle.
2) There is a bias against the side that took the prior shot taking another shot.
3) Units only shoot units in the same slots unless there are none... then they can hit anything
4) The odds of a unit shooting are prportional to the strength of the stack
5) The odds of being hit is proportionate to the size of the stack

[robin74]

1) The highest strength takes the first battle.
2) There is a bias against the side that took the prior shot taking another shot.
3) Units only shoot units in the same slots unless there are none... then they can hit anything
4) The odds of a unit shooting are prportional to the strength of the stack
5) The odds of being hit is proportionate to the size of the stack

1 and 3 are not correct.
Others are debatable in my opinion. 2 is an interesting hypothesis, though I don't think I have enough data to test it (if you took a shot, then it's more likely you'll also take the next one - but it may be because now you are relatively stronger, and not because any bias is present). I don't think I believe in 4 or 5 (I usually have a huge stack of one type and one unit each of other troops. If 4 or 5 were correct, I would hardly ever see these singles attack or be attacked. I see it quite a lot).

[ShuShu62]

1 and 3 are not correct.
Others are debatable in my opinion. 2 is an interesting hypothesis, though I don't think I have enough data to test it (if you took a shot, then it's more likely you'll also take the next one - but it may be because now you are relatively stronger, and not because any bias is present). I don't think I believe in 4 or 5 (I usually have a huge stack of one type and one unit each of other troops. If 4 or 5 were correct, I would hardly ever see these singles attack or be attacked. I see it quite a lot).

Of the five, only 5) was something that I was not taught by someone else. The other 4 have been taught to me and I can state that they have won battles for me that I would otherwise have lost. In a very real sense, the more folks you can convince I am wrong, the better off I will be.


I would appreciate screen shots of the counters to 1) and 3) as I have never seen those rules fail.

I have tested 2) over a hundred times and it has never failed, but I am leveraging the bias not relying on it solely.

4) This is recent knowledge, that has served me well to-date, 3) is a direct conclusion from 4) so a counter example to 3) would change my statement to bias from proportion.

5) I believe you are describing the fodder strategy. (e.g. 300 infantry, 1 snow pony, snow pony is hit) That your fodder is being hit instead of your main stack is not a disproof of 5), it is likely a confirmation of 4).

[robin74]

I would appreciate screen shots of the counters to 1) and 3) as I have never seen those rules fail.

I don't have a screenshot of 1) (and even if I had - how could you tell it's the first attack?), but I did see it happen (and I'm not relying on my memory here - I did make extensive notes of what happens during the battle, or else I wouldn't be able to test anything).
The screenshot of 3 is below. Both sides have all types of troops, yet archers attack riflemen.

[robin74]

I have tested 2) over a hundred times and it has never failed

Maybe I'm misreading what you're saying in 2)
I have not seen any evidence of that. If you say you're leveraging - then I would say what you do is the reason for the perceived bias. To test the existence of bias I would keep all things as constant as possible and try to quantify the probability of attack. I would be interested in learning what you mean by testing. I'm not saying there isn't a bias, but I haven't seen any evidence of it.

5) I believe you are describing the fodder strategy. (e.g. 300 infantry, 1 snow pony, snow pony is hit) That your fodder is being hit instead of your main stack is not a disproof of 5), it is likely a confirmation of 4).

Then maybe I'm misreading your 5). Or 4).
To give an example - here's how I see what you write there:
If one side has two stacks of 100 and 300 and the other two stacks of 50 and 200, all melee, then - if the first side gets an attack - the chances that the 300 stack will attack are 75% (300 of 400), and the chances that the larger stack will be attacked on the defensive side are 80% (200 of 250).
The 75% part - I haven't tested it, but anectodal evidence I have suggests it's not what happens.
The 80% part - that's certainly not what I observe. I understand you say my single cavalries will be attacked, because once cavalry is chosen there are no big stacks to be attacked. But my single melee are attacked as well, and they shouldn't (at least not as often as they are).

[robin74]

Here's what I believe about the battles.
a) Who gets the attack is random, with the stronger side being more likely to attack.
b) These probabilities are monotone in relative strength, but are not proportional to strength (there is some concave transforming function - if you have 60% of total strength, your chances of getting an attack are higher than 60%).
c) There is no inherent correlation between the rounds. You should observe weak positive correlation because you don't keep things constant (if you attack now, you are more likely to attack in the next round, because you are now relatively stronger due to losses you inflicted). Observed negative correlation is a result of stance flipping - you don't get the next attack _because_ you fortified, thus decreasing your strength.
d) Damage inflicted is connected to the _total_ strength of both sides (so as you win, the damage you inflict will steadily go down - because the sum of strengths goes down). That is of course modified by the stance of the defending unit.
e) The damage applied is lower if the attack would otherwise completely destroy a given stack - if you destroy 3 archers in a large stack, you will normally not be able to take down the whole stack of 3 archers only.
f) I don't know how a stack is selected to be chosen as an attacker or a defender. It may have something to do with the strength of stacks, I suppose it has also something to do with the number of stacks, or possible number of units. Or maybe it's something which is not modified by stance or weather.
g) An attacking stack tends to attack a stack of the same type - if available. However, that's not a 100-percent probability thing, I've seen ranged attack melee with ranged defensive units available and melee attack ranged with melee available.
h) Slider movement doesn't happen at a constant rate. It's related to either the strength of an attacking side or to the damage inflicted. Even if the slider moves in one direction during the whole battle, it can take anywhere between 24 (and even fewer with barbarians) and 49 (and possibly more) rounds to push all the way up.
i) It used to be that the slider moved for the side with the higher strength. It was certainly changed under the previous system, and I think it wasn't fully reverted back. I didn't observe enough battles after the recent change, but I think I did see it either not move at all or move for the weaker side if the weaker side got the attack.

[ShuShu62]

Maybe I'm misreading what you're saying in 2)
I have not seen any evidence of that. If you say you're leveraging - then I would say what you do is the reason for the perceived bias. To test the existence of bias I would keep all things as constant as possible and try to quantify the probability of attack. I would be interested in learning what you mean by testing. I'm not saying there isn't a bias, but I haven't seen any evidence of it.

It may be the leveraging (promised not to explain) that supplies the certainty, but I believe the shot trading is to close to actual percentages for a bias not to be present, and that is what makes the leveraging predictable.

Then maybe I'm misreading your 5). Or 4).
To give an example - here's how I see what you write there:
If one side has two stacks of 100 and 300 and the other two stacks of 50 and 200, all melee, then - if the first side gets an attack - the chances that the 300 stack will attack are 75% (300 of 400), and the chances that the larger stack will be attacked on the defensive side are 80% (200 of 250).
The 75% part - I haven't tested it, but anectodal evidence I have suggests it's not what happens.
The 80% part - that's certainly not what I observe. I understand you say my single cavalries will be attacked, because once cavalry is chosen there are no big stacks to be attacked. But my single melee are attacked as well, and they shouldn't (at least not as often as they are).

Will you agree that there is bias to the larger stacks?


As for the screenshot, I am wondering if it is limited to that exact combination... as all other archery units became riflemen so the game treats horse archers as riflemen?

[robin74]

Will you agree that there is bias to the larger stacks?

Bias relative to what?
My feeling is that small stacks are attacked more often then their relative strength would suggest.

As for the screenshot, I am wondering if it is limited to that exact combination... as all other archery units became riflemen so the game treats horse archers as riflemen?

But why would it? Horse archers themselves don't upgrade to riflemen.

[robin74]

Here's a link to some old post with a similar screenshot by Kalis - artillery attacking riflemen.

[ShuShu62]

Here's a link to some old post with a similar screenshot by Kalis - artillery attacking riflemen.

I can't argue with the evidence. The interesting thing is, the evidence, though contradicting my assumptions is confirming what I am saying.

He is complaing about only hitting fortified rifles. What I have been saying is that Kalis's position would only hit rifles (with only the occassional shot elsewhere) He is also saying that only his rifles are getting hit. That does contradict what I am saying, as I would expect kalis's ships to get pounded.

I have not seen that behavior personally, but I have only been using the knowlege of the eldar for a week, so that isn't long enough to be confident.

[robin74]

When I'm done with the current game, I will try to have a closer look at how the stacks are selected for combat. But it may take a while before I'm satisfied with the amount of data to draw any conclusions. I just hope they won't change the battle mechanics again before that

[ShuShu62]

I trust what ever you find will be correct... until they make a change, which I assume they do randomly just to encourage a 'war is hell' atmosphere.

[meagain]

1) The highest strength takes the first battle.

You mean the has higher chance of striking first. Having the highest combat stregth does not guarantee first strike unless there is a bug that causes this not to happen. I have seen no mention of such a bug.

2) There is a bias against the side that took the prior shot taking another shot.

Sadly the bias does not always come out. Having been the stronger army and seen us attacked 11 times straight before we got a single shot shows this bias does not carry much weight. it has not been a once off.

3) Units only shoot units in the same slots unless there are none... then they can hit anything

Once again I have observed units in ranged attacking melee units even though all four slot types had units in them. So perhaps once again this means higher chance of striking the same type of units.

[robin74]

So I wonder. We have seen ranged attacking melee (when opposing ranged units were available) and ships attacking melee (with ships available). Did we see any other type of troops being attacked outside their category?

[ShuShu62]

I am more curious about why some of us have seen common cross type hits and common first shot going to lesser value, and others have never seen it.

We are back to the beer bet. I do not disagree with you guys, but the coin they are flipping in my bar is not the coin they are flipping in yours. I just need to keep an eye open for that coin changing. To everybody else, you need to asses for yourself whether they are flipping Robin's true coins or ShuShu's biased coins in your bars. Ordinallrilly I would strongly recommend believeing in Robin's true coin, however since they are flipping a biased coin in my bar the best I can do is say its either the true coin, or it isn't...

[robin74]

I am more curious about why some of us have seen common cross type hits and common first shot going to lesser value

I don't think anybody claims these things to be common.

[ShuShu62]

Common enough to be relevant.

It is possible that I might win the lottery someday, but the odds are so small I live my life assuming that it will never happen.

I don't really care whether the claims I made are 100% true or just observed because they are 99% true. Either way, it would be foolish to change my behavior.

I was a physics major in college, I don't believe in 100% to begin with...

[random user]

One other aspect to consider is that this isn't a set codebase. So it's quite possible that small anomalies which appear may be caused by a bug or change in the code, which was then later reverted.

Just tossing in a random comment there. I myself have never seen cross slot hits when same slot units are available, but I also believe robin74's observation.

[ShuShu62]

...Just tossing in a random comment there....

Random comments from a random user are always welcome. Especially by someone who has majored in Physics.

What you described is exactly what I was thinking when I said I was curious about why there are varying observations of the truth.

[ShuShu62]

Funny.

I was in a 5 hour battle last night. I observed both behaviors (lost the first shot despite greater strength, horse archers hitting riflemen).

This game has been so full of bugs
--- Bloat civ swallowed all but one civ (me)
--- Civ with ghost techs
--- barbarians attacking my civ no matter what civ they are actually attacking
--- defeated barbarians still pillaging a civ
--- Medals going to members of another civ
--- Two terrains in city

That I am inclined to say that the observations are more bugs.
In five hours, the cross attack happened once, and we certainly leveraged the assumption that cross attacks don't happen to good effect. (made Firaxis 40 real dollars in one battle,

[jazzwerewolf]

I saw melee attack mobile today when there was a full range of melee troops to hit. I think that cross shot was the only one in a 5 hour battle. Generally they seem to hit troops of the same type if available.

I also noticed in another battle (yesterday) there was a cross shot too. Firstly, range hit the only range opponent, then another range attack hit mobile. I wondered if that was because the same unit stack can't be hit twice in a row?

[Adam Barrett]

no, ive seen plent of wars with the same stack getting hit multiple times in a row. generally it seems the biggest stacks get hit the most.

[Kalistes]

In five hours, the cross attack happened once

Based on a battle I just completed a few moments ago, I saw the cross-attack happen once in 53 rounds of combat, about 1.8% of the time. This was the first time I counted each round to try and get a decent sampling. Because this battle was so short, I believe that the sample size for determining the oddity of this circumstance is too small to nail this down, but I agree with you that the bias is very heavily in favor of like unit on like unit combat.

[random user]

I've seen the cross slot hit several times now, after never seen it previous to a few days ago.

I believe that there may have been a recent code change which allows for cross hits in certain situations, since there was something in common in all the times I saw it.

[AngleWyrm]

The coin flipping question needs a bit of clarity. No matter what has come before, the chances are 50/50 for the next flip. For example, the chance of flipping three heads in a row is 1 out of 8 possible outcomes, or 12.5%. But given a position of having already seen two heads, we have pruned down those 8 outcomes to just two possible future states, head-head-head and head-head-tail, both equally likely.

[ShuShu62]

The coin flipping question needs a bit of clarity. No matter what has come before, the chances are 50/50 for the next flip. For example, the chance of flipping three heads in a row is 1 out of 8 possible outcomes, or 12.5%. But given a position of having already seen two heads, we have pruned down those 8 outcomes to just two possible future states, head-head-head and head-head-tail, both equally likely.

Correct, assuming a true coin. The point about the bar bet and flipping combat is that the coin might not be true. If the coin has a bias, the indication is the bias is towards heads, so you should always bet on whatever you have seen the most of. Folks actually used that to beat roulette tables, so vegas started swapping out the roulette wheels before the bias could be worked out.

[Moonlghtknt]

I would appreciate screen shots of the counters to 1) and 3) as I have never seen those rules fail.

I've seen rule 1 fail. I don't have a screen shot, but I've seen an army that had a higher number take the first shot to a heroic stack.

I think the fact that the stack was heroic may have had something to do with it, and it might have been also that without being on heroic that side would have had a lower power total.

Next time I see something like that I'll have to take a screeny.

Rule 3 is not SUPPOSED to fail, but ever since they reverted the battle system I've seen the occasional attack on units that were unopposed by units in other slots. It's a rare occurrence, but it has happened. Personally I think this is a bug and not intentional.

[ShuShu62]

I agree that cross shots appear to be happening more frequently now than it used to, but still not enough to worry about. I have a sneaking suspicion is that you can get hit by a cross shot if you have empty slots, but that is just a theory.

I no longer think cross shots result from bugs, but rather from some algorythm I do not fully understand. I believe the opening shot observation is a bug resulting from timing (i.e. the superior strength changed in the final seconds on the screen, but the server had already determined the shot).

[Matt West]

If the numbers are pretty much equal, then everytime its a 50/50 chance for each hand/side to attack

Doesn't mean its going to be go that way, the player will win 3 in a row, then dealer 2,1,3,2,1,3 and so on :P

If you fortify you cut down the chances (by what I dunno if its exponential, I do know no matter what there must be a 5%-10% limit for the low) of getting the attack off. But yea you will save guys :P

I just kinda gauge how many people Im willing to risk losing for the bar to keep momentum/if they have a good advantage wait and d up till you get closer to increase odds

[ShuShu62]

Matt, I am not saying this to be boastful, but a lot of what I have posted in this thread is based on reproducible behaviors observed through multiple hours of battling against your teammates last weekend.

I imagine they share your belief, which is why they were slow to adapt to what I was doing. You can ask them if we traded shots because they fortified or if they fortified because we taded shots for an hour before they realized they better fortify.


There was a kid in my dorm who claimed backgammon was a game of luck. I insisted it wasn't, especially when using the doubling cube. I agreed to play him for a penny a game.

He owed me $167 after three months but still believed backgammon was a game of luck.

[KainS]

From my personal observations, I'd say the odds of unit hitting same type is somewhere between 80 and 95%, with at least a 5% chance of hitting a different type each round. Thus while it's rare, there is a chance each round that melee will hit something other than melee, but more often than not will hit melee, and the same goes for ranged, mobile, and naval.

[Monqi]

Is there any evidence that some variation of Shushu's rule 5 has any bearing on this scenario? It would explain every cross shot that I ever saw (not many, admittedly).

[robin74]

Actually, from whatever few tests I ran of this, I would say that the odds of a stack being selected (both on the offensive and defensive side) are proportional to the size (ie number of units) of a given stack.
As for cross shots, I have observed way to few of these to conclude anything, and I'd say that the odds of them happening are way lower 5-20%.

[ShuShu62]

I figured I would let you answer it first Robin before I chimed in and you didn't let me down.

I have seen so few cross shots, that I feel confident calling it a bug when it occurs.

I would even go so far as to propose cross shots will only happen in games that have witnessed the merge bug or whatever the merge bug has morphed into now that the same person can be king of multiple civs at the same time. You will know if all manner of strange occurrances are happening in the game already.