City flips - some tests and a look at the rules

[etj4Eagle]

As I have mentioned in another thread I have a strong suspicion that certain factors, including troops, effect the chance of a city flipping both ways. In the case of troops, having them in the city may only help to reduce flipping caused by resistors. But putting them in the city also causes resentment due to the presence of the military. Epecially considering that most reports (if not all) that I have read of removing troops to prevent flips have involved large garissons, this seems to be the case. And I do feel that if this is true, that this is reasonable.

Of course some of the problems with city flipping may fall back to the problem that a small portion of the community is having with combat. Ie that the random number generator is not working correctly on at least some systems, giving too many improbable numbers. In that case, many of the problems iwth the game could be simply solved by changing the random number generator.

 

[Polonius]

Originally posted by Rhandom

Polonius, insulting the people who have experienced really stupid flips by saying "Predictably enough I was pelted with rotten fruit by posters who felt certain that the AI was striking unfair low blows at them", like we were conspiracy nuts or crybabies, adds nothing.

Geez Rhandom, you really take your game problems seriously don't you!That was supposed to be a light hearted comment mildly poking fun at myself and pointing out that many people really do feel that the AI is behaving unfairly.

It seems pretty sure to me that Civ3 uses a combination of known factors (that often accumulate - as in this case) plus random elements. Unfortunately random really does mean random you can't really have "predictable random". If you can't stand randomness - despite your sig - then Civ3 is pretty much doomed to annoy you.

I'm sure that there are still aspects of the game that don't work as well as it might, but I've found it to be perfectly playable and most enjoyable. There also do seem to be ways around those times that it didn't go your way. I don't claim that the game is perfect, bug free or anything else along those lines - I just believe that it's both enjoyable and playable as it is. I enjoy the game as it is - have played it extensively, and will stop when I get bored. That's enough for me. Sorry that it enraged you so much.

 

[Northstar5757]

Polonius 100 cities and not one flip? You must be the luckiest man alive

I defend civ on just about every complaint not because I'm a fanboy but because I think civ3 is a great game. I like corruption battle results and just about anyother controversial subject. I still havn't quite decided how i feel about culture flips but here is one complaint i have.

I always have the strongest culture in my games and can usually hold most captured cities with it but some cities seem destined to flip. These tend to be ex capitals but not always. I can link them with six luxuries and rush build temples and cathedrals but they just revert back within 5-10 turns unless i wipe out the whole civ (which obviously is a very hard task). I figure I should just raze them but i really hate doing that since they usually have wonders which I like to keep even if they are obsolete.

I just think it's ridiculous that with a better goverment, luxuries, (I've noticed units seem to only stop resistors) culture, and what not I can not keep these cities.

I've decided for now I'll just give these cities away to weaker civs. They can deal with the crap

 

[Anglophile]

I too have never had a city flip, but then I've hardly ever captured any so it doesn't mean much. As to the comments that there is no warning - why should there be. You have just captured a city full of people who hate and resent you; of course they are considering bolting back to the Father/Mother/WhateverLand ASAP - unless you make their new lives much better than the serfery they have being suffering before (and even then some patriots will still go to the wall to be serfs again). I think this is actually a reasonable way to model unrest in conquered areas (think French Resistance, the Russian Resistance, the Warsaw Uprising, etc).

Culture is a critical factor and so is government. One of the fields in the editor - and so in the game - is a modifier for the 'relative government effect'; i.e. a more advanced government of the conqueror makes reversion less likely, a less advanced more likely. Very reasonable and can be modified but I for one don't really know what the values in the field actually mean.

As for the effect of garrisons, I have a couple of theories on some of the strange effects being discussed. The first is that one must hit diminishing, possibly negative returns from just piling in more garrison troops. Just as the effects of garrisons are limited inyour home cities, I don't see why they wouldn't be in captured cities as well - although the limit would not necessarily be the same. The other thought I have is pure supposition and follows the spearman beats tank threads. Every time a spearman beats a tank in someone's game, it seems to get reported as a critical flaw in the game. However, the 18 bazillion times the tank has won have not been reported. People react more strongly to the unexpected and the more unexpected, the stronger the reaction. So when moving the garrison out prevents reversion, there is a strong reaction that the game is broken. Here is my alternative explanation, based on the purest speculation without a thread of supporting data - that is why my company - Australian based by the way Polonius - pays me big bucks as the senior scientist in the Americas. At some point, we have come to realize that the computer locks in, normally for one turn, certain results. Say it does this for reversion at the end turn point. Say you have a major garrison in the city of interest. Say the reversion probability is 5% and the computer rolls a 4 - reversion. In frustration, you reload and move out the garrison to prevent their loss. Now the reversion probability is say 50% but the computer rolls 52 and the city stays in your hands. An apparent paradox and much gnashing of teeth and posts to the nearest Civ3 thread. Don't knowif this is how it works but ut is a viable explanation of the effect. The earlier point about the spearman effect is because all those times someone moved the garrison out and got reversion, they got the expected result and didn't report it.

Unfortunately, Polonius is right - unless we get the source code we will only be able to quantify the effects of the various parameters by generating huge quantities of results which can be analyzed in a statistically significant fashion.

 

[Greadius]

I've found culture a huge help in preventing flips too. I've started a new policy in how take enemy cities: I raze the nearest one to it. Usually when I go to war I have a specific territory I want and will stop there regardless; now I just have to extend that territory (much easier now that I have bombers).

In concurrent warfare, if a city flips it doesn't bother me. Or if it is an isolated colony surounded by another culture; its understandable. What bothers me are the cities that peacefully sit there on a border and then randomly flip; that is where it steps over the unrealistic boundry. There NEEDS to be a warning. A city can't collude to switch its governmental alliegence in an instant.

 

[Hurricane]

I see two major problems that should be investigated in more detail before we can draw any conclusions about the rules concerning city defection.

 Sometimes a city does not flip if garrison is moved out of city.

Different explanations have been given for this paradox (some kind of inverted military police rule, changed random seed), but I want to focus on Rhandom´s suggestion:

However... considering the bass ackwards way the game does random numbers, it is possible that changing the garison used up (or more likely, never got to - ) a "bad" "random" number on quelling resister checks, and the game then did the culture flip test after the resistance quelling check.

So, imagine the next two random rolls are 51 and 01 (scale 1-100). If you have units in the city, a random number is used to determine how many resistors are quelled. So the number 51 is used up. Next, the computer checks for flipping, and picks the roll 01, which results in a flip. Now, if the player moves the units out of the city, no roll is needed for determining number of resisters quelled, and 51 is instead used to determine flipping. This time the city doesn´t flip.

 Very old cities with much culture flipping far away from the enemy.

One fact has been neglected in this debate: propaganda. A democracy is supposed to be immune to propaganda, but what if this is broken in some kind of way? Have you tried to expose enemy spies after losing a city this way?

 

[ainwood]

I too have had the "city flips with garrison, but doesn't when garrison moves out" effect. The difference is that after a reload, I move the garrison back in about 10 turns later, and it flips the next turn. Conclusive? Probably not....

BUT: Why don't we approach this in a more scientific way. Next time this happens to someone, how about they reload, and instead of moving all troops out, just move out 1. Then move out two, then three etc. See if there is a critical level.

I think that this is the basis of the gripes for flipping of captured cities - its not so much that it flips, its that it takes most of your army with it!

 

[eyrei]

The only cities I ever have defect are ones I conquered and have foreign nationals as citizens. My reactionis simple. The first time I capture a city, I try to hold it. If it flips, I send in some more units, and raze it. Sometimes I just assume a city will revert. For instance, if I take a city on an island that is completely controlled by another civ, and is reasonably close to the cultural center, I just raze it and move on. Cultural flips during war by a city that is still in resistance, or is filled with foreign nationals should be attributed to partisan military actions, and not to some sort of peaceful capitulation by your governor. I think this effect was added to make conquest of other civs more difficult than the unit vs. unit combat that is animated for out benifit. Imagine the level of resistance if a foreign army captured New York city, and then tried to use it as a launch pad for further invasion. An overwhelming assault on a coastline where the majority of cities in the area are taken withing a couple turns is a different story, and this is to some extent represented in the game. Normally, if I take the cities surrounding the city in question, so it does not really have a border with its motherland, then the city remains mine. I really think those people who have such a huge problem with cultural flips need to reasses their strategy. I think these miniture revolts are one of the best features of CIV3, as they add a new level of difficulty to conquest.

 

[Richard III]

Originally posted by Polonius
Predictably enough I was pelted with rotten fruit by posters who felt certain that the AI was striking unfair low blows at them, and that I didn’t know what I was talking about.

For the record, as one of the fruit-pelters, I didn't take this as unfair. While I'd like to hear more from the Gods at Firaxis, we should at least give Polonius credit for trying to have a calm, reasonable exploration of this. Like Picker in "Primary Colours," asking "Americans to calm down a little" and have a "conversation" about our future. His search through the manual, etc. is certainly more careful than mine...

Polonius, all that said, do you still beleive that all troops in an occupying garrison should flip with the city? I've been thinking about this in the context of the historical examples I tossed out elsewhere, and I actually agree with this if the flip is original (e.g. from a never-conquered city) or if the flip takes place over 100 years, say, after a conquest, since the garrison by then would be "local." Otherwise, complete destruction of whole occupying armies still seems odd to me.


R.III

 

[CurtSibling]

Originally posted by Mack the Knife
Polonius, i know you guys have nice beaches, pretty girls and weird animals over there. you need to go outside more!

Well, if it stops weekend warriors whining about the game being unfair, I'm with him!

Read the manual, you guys!

 

[Peteus]

Ainwood, That's one gripe. Here's a summary of most of my gripes:

- The feature is very poorly documented. What factors influence culture flipping? If a large garrison makes a city more likely to flip, tell us! Then we would be able to do something about it. Don't give us a scattered list with parts missing and parts wrong, and tell us to fill in the gaps ourselves!

- No warning when we are doing something dangerous. If the city was mad about the garrison, there would be pelting with fruit, etc. And they wouldn't all be happy!

- Cities flip when they are far happier than they were with their old civ. How does that make any sense?!

- Cities flip when the civ they are flipping to has far less culture than the city they are flipping from. (I'm talking 2:1, not 4:1)

- Cities flip when you are in Democracy, so propaganda can't be a factor. Or is it broken for the human (the protection works for the AI) like Air Superiority was? That's a scary thought, and very difficult to test.

- The effect of the palace and forbidden palace are documented poorly, and probably inaccurately.

- Garrison perishes, when it is strong enough to raze the entire civ on its own!

- The knife doesn't cut both ways, because if the AI is capturing your core cities the game must be over anyway.

- Testing is almost fruitless, because of the complexity of the issue and the length of a game. If anyone has a lot of time on their hands, here's "all" you'd need to do...
-- Play at the easiest difficulty, so that the AIs don't really threaten you. Pangaea might be good, lots of land for convenience, maybe 8 civs on a normal map. Make sure some of them are high-culture sorts like the Babylonians.
-- Intentionally build up slowly, keeping your culture on a par with the other Civs' average.
-- Research Infantry/Artillery and Tanks. Build up a large armed forces and start picking off the AIs. Capture all of their core cities, but leave some scattered outposts. Treat some civs differently than others - leave a couple of them with their original capital, leave one or two with only an island outpost, etc.
Print the map, and start making notes. List all of the captured cities. Record the culture they had before you captured them. Record the distance to the old capital and the new one, and the distance to your capital and your FP.
-- Start saving at the end of each turn.
-- Place garrisons of various sizes in the different cities. Allow some to stay in revolt for a while. Record garrison sizes.
-- Rushbuild improvements in some cities, not in others. Record the number of turns from the time you captured each city to the time you built each improvement.
-- Join your workers/settlers to some cities, and third-party captured workers to others.
-- Record the happiness status and nationality of each citizen in each city every turn!!! (If you were thinking that it sounded OK so far, I bet I lost you there!) Also record the culture ratio between your civ and each of the others every turn.
-- Whenever a city flips (and I'm sure some will), revert to the previous end-of-turn save and make a note of the total civ culture ratio, your accumulated culture in that city, the citizen's happiness and nationality, the total culture and culture income of the nearest cities of yours and theirs. Try adjusting the garrison size and make notes on the results. (Just a note - don't treat it as a sample, because it would be tainted.)
-- Drag the game out as long as possible. This may require leaving one or two civs strong, to keep you from getting a domination victory too fast. (You may need to give them some of the captured cities too. Give away fringe cities which are less likely to flip and keep the core ones, maybe.) Can you turn Domination off? That would give you more leeway.
-- Analyse the results! I suggest a Cray or similar supercomputer to do a proper treatment of all of the collected data. You should be able to isolate all of the variables that people have come up with to date.

-or- we could keep pressing Firaxis to tell us how this really works instead. I'm for the latter suggestion!


Culture is a good idea, and its effect on worked tiles, borders, etc is one of the best improvements in the game. But the way cities flip needs to be dramatically revisited!

 

[eyrei]

Originally posted by Richard III



Polonius, all that said, do you still beleive that all troops in an occupying garrison should flip with the city? I've been thinking about this in the context of the historical examples I tossed out elsewhere, and I actually agree with this if the flip is original (e.g. from a never-conquered city) or if the flip takes place over 100 years, say, after a conquest, since the garrison by then would be "local." Otherwise, complete destruction of whole occupying armies still seems odd to me.


R.III

This is the only real problem I have with the programming concerning revolting cities. While it is possible the populace might defeat an occupying army in some way, it is highly unlikely they would completely destroy it. Seems like some of the units should be destroyed and maybe the rest appear at the edge of the city's cultural border with one hit point left. This would probably alleviate some of the irritation people are showing with the entire concept. It is one thing to see a conquered city revert, but it is another entirely to see that city's reversion cause your 20+ unit offensive force to simply disappear.

 

[Moulton]

I maight have a saved game where I can run some experiments. I had destroyed the Romans, and IN the process, lost one city to flipping-- my own fault there, it was resisting, and I neglected to leave a garrison. THAT city is not a test. BUt there were a number that possibly could be tested.
IN a later game I had one city flip during a conquest, where I had done all the right things.. and they were in awe of my culture. They wer Russian.
In a later game, Bab flipped one of my very own cities, early in the game. It was slow in growth, high corruption, and I was still making its initial garrison--had not built any culture. It was also two squares from their capitol. More importantly, It controlled an iron mine--THE reason for its placement. The Russian city controlled luxuries.
It is hard to escape the feeling that if the AI WANTS something badly, it bends the rules. I have seen foot units (bowmen) move two squares to attack a unit of mine... I saw 1 1hp (red) archer defeat 4 Immortals in a row.
Well, I WANTED that iron. REstart did not change anything. We know about that. REstarted and started a temple instead of the second garrison unit(spearman) At 1 shield, this was a long wait. And I did not want to risk hurrying in despotism. Anyway, the temple finished, eventually, and the city never flipped.
Even if I had not kept my city, I could see probably cause for the flip. That's ok as a game design, and not a bad model of RL.
I later dumped that game, (I was losing )
The reasonalbe chance for success or failure is OK.
The wild unpredictable flip is unreal. Hasnt happened to me, yet, but I am not holding my breath. It would be like NewYorkCity suddenly declaring itself to be French...

 

[Moulton]

Just an additional note, I spent the last 2 hours of game time last night watching the AI virtually wredking my fledgling empire. Bab attacked with hordes of Bowman. I mean, we are just starting into the middle age, and she sends hundreds--no exageration--of bowmen. I don't know how she coul dmake so many. I still have the top score, the most 'power', but I cant make that many. Most of them died against entrenched Immortals, but I still had to devote my entire economy to making Immortals, spearmen and horsemen. Then Aztecs demanded Monotheism, and attacked from the other side. Most of their attackers ended up dead, but they did take two weak cities... and
Here is the point .. one of them flipped back two turns later.

I lost only one city in the whole mess, but the cost was staggering.

 

[jju_57]

I would like to make a couple of points. It seems that some people constantly experience flips while others, myself included, have either never experienced one or had one that totally made sense. A logical conclusion can come from this. Playing style also influences flips. Playing style simply means that what you do in one game is similar to what you will do in another game. To better determine what causes a flip you need to analyze the different playing styles.

The second point is that since this is a computer program, then the flips are based on a set of code that is logic based. While I do believe some random factor modifies the input, I don't believe there is a strict random chance for any city to flip. And since the code has to be logical then a set sequence is folled by the computer to determine the flip. Therefore is IS based on something. Now the question remains if we the players should know each rule and formula in the game. Some people do and some don't.

 

[Peteus]

Now the question remains if we the players should know each rule and formula in the game. Some people do and some don't.

This seems to be a common argument from the "suck it up" crowd (not directed at jju_57 himself necessarily.)
I don't want to know the formula exactly, although it wouldn't hurt. What I want is for Firaxis to publish a complete and accurate list of the criteria for flipping. Ex. "Oh yeah, we forgot to mention that putting an over-large garrison in a city reverses the positive effect of a small garrison!" (If that is the case - who knows for sure?!) If a flip happens out of random bad luck, I want to know that it was just bad luck instead of having niggling (and completely unprovable) doubt that I was doing something to make the situation worse. A real-world leader has economists, generals and advisors telling him/her what went wrong or what is threatening. The feedback may well be biased or contradictory, but at least there is feedback! Instead, all I get is my military advisor giving me some stupid report along the lines of "The people of Stalingrad have decided to reject the enlighted way of their new, ecstatically happy lives, chosing instead to revert to the sqallor and filth of their previous existence. They will revel in their return to their now-defunct former civilization for the half-turn it takes you to swing your mighty armies around and inevitably slaughter every last one of them, corrupting their fields with the salty blood of their populace."

Since nobody can claim that it works in a real-world way, if it's "just a game" (which I completely agree with), simply tell me what the rules are! The manual and civilopedia are incomplete and often misleading, with rules that were missed completely or changed after publishing. So I want them updated to show how it really works. I feel like I'm trying to play Monopoly without the property cards!

 

[eyrei]

I would like to know for certain what factors affect cities revolting, but I don't think I need to know the exact formula. There are probably variables in the formula that we would not be able to access anyway. I do sometimes get the feeling that a large garrison does adversely affect the chances of a city flipping.

 

[jju_57]

First off, for the record I think the manual should have a chapter that shows the complete formula like the old SSI games did.

As for wanting to know what factors go into it would it really help if I said "culture, garrison, capital, happy/resitors and distance to capital"? Or do you want the formula that says start with 100, divide by population minues garrison unless garrison is population times 1.25, etc, etc. etc.

That all said and done I find it interesting that players either experience constant flips or none at all. It does go to show that playing style IS a factor.

 

[Peteus]

As for wanting to know what factors go into it would it really help if I said "culture, garrison, capital, happy/resitors and distance to capital"?

If that were all of the factors, it would be dead easy to figure out and we wouldn't be having this discussion! There are clearly many times that many variables. Several different culture effects, lg garrison vs small garrison possibly, citizen nationality, WLTKD, possibly individual happiness factors, courthouse, possibly corruption, possibly FP vs real palace (because what is the capital besides the city which houses your palace? Moving the palace moves the capital, right?), possibly starvation, mobilization, unhappiness effects possibly having a direct effect, possibly road link to the capital, etc, etc! And many of them are not properly documented, so all we can do is guess. That's what needs to be changed.

So yes, if the 5 factors you mentioned were all there was to it, I'd be content with your list! What I don't like is "these five factors have an effect, as do 15 others that we won't bother to get into.

 

[Peteus]

For those who "haven't seen any flips" or "would like to see an example of one of these super-cities flipping"...it didn't take me long to come up with one! It should work for you - oddly the random seeds were apparently preserved on my PC after a reboot - I thought it wasn't supposed to work that way? Anyway...
Alleghany was recaptured by me (Iroquois) 3-4 turns ago. (It was originally mine, then Babylonian, then Aztec a couple of thousand years ago.)
I could have razed it, but didn't. After all, it's only 11 tiles from my capital, and only 3 tiles from my FP, so I consider it rightly part of my empire!

So meet Alleghany:
- 7 Happy, 6 still Resisting, and 3 Entertainers = 16 Citizens. We all seem to be assuming that Resistors and Happy citizens affect the chance of revolt, but that's not one of the factors listed in the Civilopedia!
- It has a temple (just built - no problem because there's almost no corruption & the Iroquois are Religious), marketplace, courthouse, and police station (captured.)
- It has 6 luxury resources, for a total of 12 smilies with mktplace.
- JS Bach's Cathedral is helping contentment.
- The foreign capital is around (10 tiles away), but no closer than mine (11 tiles.)
- 4 garrisoning troops are working on surpressing the revolters. As an aside, my three Mech Inf + 1 cav fortified in a Metropolis could probably withstand the assault of the Aztec's entire remaining army!)
- I'm in Democracy, so propaganda should be impossible.
- No Aztec cities are bordering Allegheny.
- The city is not in civil disorder. (Those last two are mentioned among the five factors in the Civilopedia.)

Pretty good example of a "SuperCity" (not that I haven't witnessed better.) Load it up and watch it flip!

Oh, and I think I know what the big danger factor is...the one big possible issue that exists is that the city had accumulated 3300 culture under Aztec rule before I captured it. I suspect (completely unsubstantiated, of course!) that this is the biggest factor of all. It would explain why core cities are the most likely to flip, even when the civ has no presence left in the area.