Culture-Flipping Exposed

[duodecimal]

Originally posted by Hurry
How are population of destroyed civs counted? As your own pop or foreign? In other words, would a city of size 12, with 6 own, 5 destroyed and 1 actual foreign pop in fact count as it had 6 foreign population?

I don't think killed population is counted. In your example, the city would have a population of 7.

Someone else asked about civilizations coming back to life through a city defecting back to them - I don't think that's possible. There must be another city somewhere that you don't know about. If a city defects to a civ you thought was destroyed, then it couldn't have been destroyed. Either your foreign advisor tells you that a civ has been destroyed, or, if you miss that message, you can look for their leader on the diplomacy screen or the civ in the espionage/diplomacy windows.

It would be neat if that could happen, though.


As far as Specialists, Resisters, and any weighting differences between Happy, Content, and Unhappy laborers, I think Firaxis probably needs to clarify a bit more. The more I try to adjust the formula, the more I see problems crop up (such as negative numbers if there are a large number of resisters, etc.).

This is the kind of stuff that should be found in a strategy guide, but there haven't been any good ones that I know of since the Emerich/Hughes guides for Master of Orion and that other game (Master of Magic?). Dan mentioned a civ fanatics forum - I'll venture over to see how they're digesting Dan's post.

Duodecimal

Edit: What a bunch of whining ninnies! It's a waste of time, but here's the thread:

http://forums.civfanatics.com/showth...threadid=14337

Couple tidbits: Looks like Foreign Palace does *not* have all the benefits of the Palace, as mongoose suggests above. Also - Dan (IIRC) says that 'shortest route in tiles' (as the crow flies) is the distance method, but he isn't sure. I'd forgotten about the coordinates given in-game - tonight I'll see if there's any significant differences in ratio whether you count tiles or use sqr root(x^2+y^2).

[mengo76]

This is what Dan said in the civfanatics thread:

"I would assume this is measured in tiles, based on the pathfinding algorithm's determination of the shortest path from the city in question to the civ's capital city. It may be even simpler than that, I'm not sure."

Well I would think its is *not* simply using the pathfinding algorithm... Lets say a city is 5 squares away from your capital, but there is an inland sea in between them, such that if you were to move a unit from one to the other you would go through 8 squares... I think the distance in the formula would rather be 5 than 8. And what if the city and capital are in different continents?

But even if it were the shortest number of squares... I'm pretty sure that they use pythagoras for corruption calculations, so why wouldn't they use that for culture flipping too? My only question is, would the X and Y axes be the NE and NW diagonals, or would they be using the coordinate system you can see in the map editor? I made some calculations with the former and they seem to fit, I need to try the latter...

[Xin Yu]

I remember somebody posted that if you don't have a capital then you don't have corruption. Since corruption maybe based on distance to capital, that could mean that any city's distance to the capital is zero if you don't have one. Now come to the culture flipping thing, does that also drastically increase your chance of gaining a city? Or just the opposite? If the opposite is true then all you need to do is to destroy a civ's capital.

[rflagg]

Hello,

Yeah, DuoDecimal, I was the one who brought up the possible scenario of a Civ coming back through culture after they are wiped out. It didn't happen in my game, I was just curious if it has happened to anyone, or if it's even a possibility. My assumption is that once all of the cities of civ A are destroyed/claimed by another civ (lets call him B), that even if the culture of civ A greatly outweighed B's culture, there is no way for A to ever come back. It seems as though once a civ is destroyed, he is wiped from memory (other than the historical data), so it isn't possible for a comeback. Unfortunately, this makes me slightly sad, because I think it'd be a good idea, much like the idea in Civ 1 where attacking and taking over the capital gave the possibility of a revolution of the civ being attacked, such that it'd split into two civilizations - one civ, the rebels, would be supporting your war effort against the loyals. I'm 99% sure it's not in Civ 3, but it is a feature well missed.

Anyway, perhaps those two items can come into play in the spring with the new add-on pack. We can only hope, anyways.

-Rflagg.

[vmxa1]

Once the advisors says the civ is gone for good, it has never come back in any of my games (more than 20 now). When I kill the last city in some games, it is not the end of the civ and I do not get the msg. I will soon find another city come out of nowhere.

[barefootbadass]

Originally posted by Aeson
What I'm sure about distance...

Distance is the number of tiles between point A and B, each tile counting either 1 or 2 depending on its alignment. I'm not sure exactly how this is calculated. (X2 - X1) + (Y2 - Y1) doesn't work because of how every 3rd diagonal is treated.

Diagonals (1st,2nd,3rd..12th) = 2,2,1,2,2,1,2,2,1,2,2,1
All Verticals and Horizontals = 1

Nice work. Although I would guess that the 3rd diagonal being 1 based on corruption is because there is some benefit gained back for having your cities without overlapping city radii(did you go out that far on the horizontal? the 3rd diagonal is the first square where you have no overlap) . I would guess that it only counts moving along the vertical and horizontal, two movements in those directions would be equivalent to one on the diagonal, and that there may merely be static 'rings' where the diagonal is 1 from your capital, where your 'optimal' spacing would be.

[frankdog]

In my last game, I captured Beijing. After a few turns the resistance was stopped and I rush-built the Forbiden Palace in it.
A few turns after that, it reverted back to China.. FP and all!

It was moderately garrisoned, I thought the FP would would act like the Palace and secure it for me!

[DrSpike]

Yeah I thought that too. Fortunately I managed to stop myself breaking any peripherals after it happened..............just

[Xin Yu]

So, in order to gain advantage, I'll trade an AI civ a relatively big city (not too big since I don't want the city to starve when my other cities are pushing boarders against it), so my citizens will outnumber AI citizens. The city should have had temple, etc. for several turns so that my culture will outnumber AI culture. The city should be close to my capital and I have other cities close to it as well. I'll make sure the AI government type cannot sacrefice people for rush building. I'll pick an AI which has a lower culture than mine to make the deal. I'll remove all units from the city so the city is likely to go riot (since it's big). Now the city will flip back to me soon and I gain benefit from trading the city to the AI. Anything missing?

[duodecimal]

Just ran a test...

Originally posted by Xin Yu
So, in order to gain advantage, I'll trade an AI civ a relatively big city (not too big since I don't want the city to starve when my other cities are pushing boarders against it), so my citizens will outnumber AI citizens. The city should have had temple, etc. for several turns so that my culture will outnumber AI culture. The city should be close to my capital and I have other cities close to it as well. I'll make sure the AI government type cannot sacrefice people for rush building. I'll pick an AI which has a lower culture than mine to make the deal. I'll remove all units from the city so the city is likely to go riot (since it's big). Now the city will flip back to me soon and I gain benefit from trading the city to the AI. Anything missing?

When you give a city away, all its population become citizens of the culture you give it to.

In the test I ran, I gave away my third city, which ranked #5 on the Top 5 Cities list (I had two other cities there - my capital and 2nd cities, Atlantis and Arbomare). The city I gave away, Antilla, had built Magellan's Voyage about twenty turns ago, and, since my culture is based on the Babylonians (no editing, just renamed everything), all other cultural improvements have been built - it's the year 1400 AD. Antilla restarts with a culture of 0. Two other cities, Aurora and Solaria, are pushing hard on it (9 tiles and the three remaining tiles are not in anyone's cultural influence), and my culture beats England's about two-to-one (where can I find the exact cultural value of other cultures?

I counted tiles, did the pythagoras thing, and figured that London is 23.6 tiles away, and my capital is 4.9 tiles away. Antilla is not in civil disorder nor WLTQD, and has 10 laborers (the non-laborers are getting starved off). I have never been at war with anyone, regent-level, and the english are my only neighbors on this upside-down "U" bent-barbell-shaped continent (excellent).

I went seven turns, and no reversion. So, let's look at the numbers:

((Citizens+Tiles) * CultureRatio * DistanceRatio * StatusFactor * HistoryFactor ) - Troops

English Factor:
((10+9) * 1/2 * 4.9/23.6 * 1 * .5 ) + 1 = 1.99

Atlantean Factor:
((0+9) * 2/1 * 23.6/4.9 * 1 * 2) - 1 = 172.4

"Odds" of reversion: 98.9%

Obviously, the city should have reverted during those seven test turns, but it didn't. Is there a "breather" period of X turns when you hand a city over? Or is there a completely different formula for retaining a city (all I did was invert everything in the formula). Maybe 172.4 does mean something alone - which isn't unlikely since the factors in the formula are not going to lead to much different results whether it's 100 BC or 1920 AD. Is 172.4 a number in basis points (1/100th of a percent)? 1.72% doesn't sound like much. 17.2% would likely have led to even odds of a defection during the 7 turns (I guess - I'm not that good at statistics).

Anyhow... I'm getting into the late Industrial ages, and am yearning to start a new, clean game on a similar map - if only I can remember all the settings I used, and hopefully there'll be more than 3 coal tiles on the entire huge map. I'd disabled resource depletion in the game rules, but I guess I'll have to bump up odds-of-appearance a little - there seems to be almost enough iron, so I'll make coal almost equal to that.

Duodecimal

[Sirian]

What about Courthouses? How do they factor in flipping? The docs say they make a difference, but no mention in this explanation. So which is the truth? What's the deal?

- Sirian

[DrSpike]

I think they help against propoganda based defections, but not reversions.

[Murtin]

Dan's post clarified a lot. For instance, it should be impossible for a city to defect if it has no foreign nationals and no foreign culture occupying its city radius. Anyone got an example to prove that assumption wrong?

However, there are still lots of info missing. Trying to construct a formula from what was said is probably a lost cause. First, remember that there are unknown weights attached to those factors that are applied multiplicatively. And though we know that relative distance to respective capitals is more important than culture ratio, we don't know just how much more important it is.

Second, even if we knew those weights, we don't know what to do with the number computed. Somehow it is to be transformed to the range [0, 1] and be treated as a probability of defection, but there's no indication of how.

Originally posted by duodecimal
"Odds" of reversion: 98.9%

Obviously, the city should have reverted during those seven test turns, but it didn't.

Yeah, this is a sure sign the probabilities aren't computed quite like that. A 1.1% chance of NOT defecting in one turn, means the chance of not defecting in seven turns is less than 2 x 10^-14.

Is 172.4 a number in basis points (1/100th of a percent)? 1.72% doesn't sound like much. 17.2% would likely have led to even odds of a defection during the 7 turns (I guess - I'm not that good at statistics).

With a 17.2% per turn chance of defection, the probability of defection within those seven turns would be
1 - (1 - 0.172)^7 = 0.733.

[Flight]

ignore

[Unregistered]

Originally posted by Dan Magaha FIRAXIS

The base values used to determine the chance of city flipping are as follows:

A) The number of foreign nationals in the city in question (resisters are counted twice), and

B) The number of the 21-tile city-radius squares of the city in question that fall inside your cultural borders.


These numbers are then further modified by a variety of factors, applied multiplicatively. Here those are, in order of importance:

{snip}

6) Lastly, the number of land-based combat units (e.g., any unit with at least 1 point of offensive and defensive capability) in the city in question are subtracted. This factor is relatively low on the totem pole and this shows you why cities can flip even with huge militias garrisoned in them.


Hope this helps.


Dan

Has anybody ever had a city flip that had a garrison of offensive/defensive capable land units greater than the city pop? Or greater than the foreigners pop?

I haven't had any trouble w/ cities flipping since I began leaving at least as many off/def units as there are pop points in the captured city. And some the the cities involved have had as little as six squares that belonged to me and as many as 16 pop, all foreign nationals. That particular city was definately closer to the foreign capital than my own and the other factors were pretty much a wash, but after putting 16 units on garrison duty, it never did another flip.

Thanks,

[DrSpike]

Originally posted by Unregistered


Has anybody ever had a city flip that had a garrison of offensive/defensive capable land units greater than the city pop? Or greater than the foreigners pop?

Yes, and yes.

I have had some strange results as well, like a captured city reverting when my capital was closer, they were no squares not in my territory, my culture was greater overall, and I had reached 100 culture points myself in the city. Also, as you report I have had cities hold on for ages when everything in the formula suggests reversion would occur quickly.

[Unregistered]

Originally posted by DrSpike


Yes, and yes.

I have had some strange results as well, like a captured city reverting when my capital was closer, they were no squares not in my territory, my culture was greater overall, and I had reached 100 culture points myself in the city. Also, as you report I have had cities hold on for ages when everything in the formula suggests reversion would occur quickly.

Well, I just had Moscow flip back to the russians last nite. It was three times closer to their capital, I had 15 off/def units on garrison duty, the city had 24 pop, 12 of them russian, I had accumulated around 700 culture - but since it was their former capitol I'm sure the russian culture was much greater than that - and there were 4 unhappy pop pts, so no wltkd. Reloaded & placed 4 more off/def units on garrison duty & voila, no flip.

learning continues, thanks for the response.

[DrSpike]

That is interesting, especially if you only went back 1 turn. It seems the advent of the formula will do little to reduce confusion in this area.

[Murtin]

Originally posted by DrSpike
That is interesting, especially if you only went back 1 turn. It seems the advent of the formula will do little to reduce confusion in this area.

Yes. We don't really have much of a formula yet, though. And even if we knew of the exact formulation, the city-flipping would still depend on random numbers, so over a large number of games, involving a large number of cities and a large number of turns, very improbable city flips would still occur every now and then.

Now I'm gonna bore everyone to death with even more formalized speculations.

Definitions:
n = # of non-resisting foreign nationals in city,
r = # of resisting foreign nationals,
t = # of city radius tiles occupied by foreign culture,
p = ratio of distances to respective capitals,
a = weight factor for distance ratio,
q = ratio of cultures,
b = weight factor for culture ratio,
m = city culture memory factor,
s = Disorder/WLTK factor,
u = # of units with A/D values garrisoned in city,
c = weight factor for number of units.

Then I guess the formula would look like
(n + 2r + t) ap bq m s - cu, with a > b > c > 0.

But this looks like it would break Dan's claim that civil disorder or higher city culture memory of the attacking civ will double the chance of a flip, or that WLTKD would cut it in half. The subtraction in the end would spoil that assertion. So maybe it's like this instead:
(n + 2r + t - cu) ap bq m s, with c << 1.

This would enable factors m and s to neatly double the chance. And now you would "only" need to know the value of c in order to know just how big a garrison it will take to eliminate the risk of a flip.

Then there's the question if a, b and c are tuned so that the raw number thus computed can be used directly as a flip probability, or if the raw number is further transformed...

[kittenOFchaos]

I've got a suspicion that the equation is wrong and that military units in the city PROMOTE flipping...

[Unregistered]

Originally posted by kittenOFchaos
I've got a suspicion that the equation is wrong and that military units in the city PROMOTE flipping...

Well, I think that the calculation/code doesn't handle the case of no units on garrison duty correctly. One of my original solutions to the flipping problem (no pun intended ) was to just reload & move the entire garrison outside the city walls and then end the turn again. My original intent here was to just retake the city on the next turn after it flipped, but I never had a case where this didn't prevent the flip on the next turn. However, I view this solution as a pure exploit of the game mechanics & I didn't like doing it. One would think that if a city of size 12 was going to flip with 8 off/def units on garrison duty then it would still flip in the following turn if there are NO units in the city!

One grain of salt here, my sample size is kinda small, so ymmv. I don't think I took advantage of this exploit more than 6 or 7 times before reading the transcript of one of the chats here on poly where Dan or Soren made several comments on how to prevent flipping. & those comments were "wltkd helps", "garrisoning as many units as there are population points helps" and "catapults & cannon don't count as units for this". It's nice to have it laid out a little clearer, thanks Dan!

c ya,

[Murtin]

I've got a suspicion that the equation is wrong and that military units in the city PROMOTE flipping...

Well, I think that the calculation/code doesn't handle the case of no units on garrison duty correctly.

I wouldn't be surprised if the actual equation turned out to contain some wicked non-linearity... and Dan's post accordingly was just a bunch of simplified approximations.

[Dodgy Geezer]

Originally posted by Unregistered


One of my original solutions to the flipping problem (no pun intended ) was to just reload & move the entire garrison outside the city walls and then end the turn again. My original intent here was to just retake the city on the next turn after it flipped, but I never had a case where this didn't prevent the flip on the next turn. However, I view this solution as a pure exploit of the game mechanics & I didn't like doing it.

My take on this is to only keep cities that have wonders, and then to use a minimal garrison inside, and what would have been the garrison just outside. That way if it flips I simply retake it. Eventually the city seems to get the message, or I get bored and raze it. I have never had a city flip more than twice, but of course I am normally relocating the opponents palace further and further away as time goes by.

And remember, only sure way to stop a flip is to remove the civilisation that it belonged to.

[Dodgy Geezer]

Originally posted by Xin Yu
So, in order to gain advantage, I'll trade an AI civ a relatively big city (not too big since I don't want the city to starve when my other cities are pushing boarders against it), so my citizens will outnumber AI citizens. The city should have had temple, etc. for several turns so that my culture will outnumber AI culture. The city should be close to my capital and I have other cities close to it as well. I'll make sure the AI government type cannot sacrefice people for rush building. I'll pick an AI which has a lower culture than mine to make the deal. I'll remove all units from the city so the city is likely to go riot (since it's big). Now the city will flip back to me soon and I gain benefit from trading the city to the AI. Anything missing?

With the 1.16 patch you can't trade a city for anything. All you would gain from the flip was the single defensive unit created in the city at time of flip. There was some discussion on another thread as to whether this could be exploited as a way of creating defensive units.

Also, you don' t need to remove your units, they will be sent to your capital when you give it away, which incidentally I have used to regarrison my capital on one occasion when there was no alternative.

Someone has already dealt with the change of citizenship point, but presumably its culture doesn't get transfered, so there is still a chance of a flip back.

[Xin Yu]

quote:

Also, you don' t need to remove your units, they will be sent to your capital when you give it away, which incidentally I have used to regarrison my capital on one occasion when there was no alternative.

I borrowed a CD from a friend and when the patch was out I had already returned the CD. Didn't have a chance to try it. (uh, yes, I still haven't purchased it). Now, what happens if I have ships in a given-out city while my capital is not a shore city? Do they give me inland ships?

[Soren Johnson]

Originally posted by Xin Yu
quote:

Also, you don' t need to remove your units, they will be sent to your capital when you give it away, which incidentally I have used to regarrison my capital on one occasion when there was no alternative.

I borrowed a CD from a friend and when the patch was out I had already returned the CD. Didn't have a chance to try it. (uh, yes, I still haven't purchased it). Now, what happens if I have ships in a given-out city while my capital is not a shore city? Do they give me inland ships?

I believe they will be in the nearest coast (not lake) square.

[Unregistered]

Originally posted by Dodgy Geezer


With the 1.16 patch you can't trade a city for anything.

Well, I've been able to get cities during peace negotiations w/ the patch. But you're right, I can't get them to offer me anything for a city I'm trying to give away. In fact, if I give a civ a city as an outright gift, even one that had been theirs in the recent past that I've just liberated from another civ, it seems to piss them off. I should say that I applied the patch mid-game, & I've heard that can muck some things up, maybe this is one of those things.

Cheers,

[Dodgy Geezer]

Originally posted by Unregistered


Well, I've been able to get cities during peace negotiations w/ the patch. But you're right, I can't get them to offer me anything for a city I'm trying to give away. In fact, if I give a civ a city as an outright gift, even one that had been theirs in the recent past that I've just liberated from another civ, it seems to piss them off. I should say that I applied the patch mid-game, & I've heard that can muck some things up, maybe this is one of those things.

Cheers,

Sorry, I will clarify my comment. You can' t SELLl one of your own cities to the AI for anything - you can only give them away post 1.16 patch. Yes, you can obtain cities from the AI, which incidentially is a good way to obtain cities with a much lower chance of flipping as they are all your own citizens.

I have never noticed the AI being unhappy about receiving a city, but then I seem to spend most of the game with the AI furious with me = must be my play style! Either way, they don't seem to refuse gifts

[Easy Rhino]

dumb little comment regarding probability.

Remember, if the chance of flipping is "only" 5% or so, that's still checked EVERY SINGLE TURN, so the odds of a flip happening EVENTUALLY are pretty decent.

ER

[dadacp]

As others have said, it's possible to have no capital by building an attack force with the starting city and and disbanding the starting city before conquering and keeping an AI city. This has the immense benefit of all cities having little or no corruption. It appears as if the distance to the capital is always zero when there is no capital. Does this also mean that cities won't flip back to the AI?