The Curious Cat - City Upkeep Explained

Well the original tests were on cities of size 1, no culture
So far it looks like
City Distance maintenance may depend on
Distance
Population of city
Cultural level
Number of cities? seems redundant

City Number maintenance may depend on
Number of cities
Population of city
Cultural level
Number of this city (age/distance from palace..to avoid rounding errors)
 
Krikkitone said:
Well the original tests were on cities of size 1, no culture
So far it looks like
City Distance maintenance may depend on
Distance
Population of city
Cultural level
Number of cities? seems redundant

City Number maintenance may depend on
Number of cities
Population of city
Cultural level
Number of this city (age/distance from palace..to avoid rounding errors)

As I said, my city had that distance maintenance of 1 since it was founded (and it was my second city). And my other city didn't have the distance maintenance even though it's 6 tiles away (farther than every other city). OTOH, all my other cities behaved as you would expect, it's just those cities which were weird....

Req
 
I am sorry to hear that we may be cursed with the RCP strategy again. I can't believe that Firaxis intended this especially since the Conquests expansion to Civ3 specifically eliminated it. The problem with RCP is not that it enables you to eliminate some of the disabilities that a city experiences as it get further from the capital, but that in doing so it forces you to play the game in an artificial manner. You have to slavishly place your cities on a particular ring instead of where they should be located in order to take advantage of the terrain and resources available on your map.

It encourages participants in competitions such as the GOTM to "play the rings" because then they can increase their score.

I realize these experiments are preliminary, but I hope they don't turn out to be correct.
 
Ok, there's clearly more going on than I thought. I'm not sure when I'll get time to run more in-depth tests, as my machine is still painfully slow even with the patch, and I'm working on the GOTM at the moment. From my experience playing, I've noticed the same thing, that distance upkeep seems to work out higher in practice than in the small-city tests. I've observed for a fact that there appears to be a scalar difficulty factor that gets multiplied in before rounding. There may be other multipliers as well. Any observations would be appreciated.

@Thallasicus
Yes, it seems that the AI still does go for the frozen seafood aisle on the map. Then again, so do I after some consideration. Even a bare water tile can give you 2 commerce when worked. So plop down a fishing town and you can fairly easily get:

+3 commerce from trade routes ( if you have currency + corporation)
+1 gold from shrine income per holy city you own
+3 or +6 commerce equivalent if you run mercantilism
+2 commerce from the city tile itself
+2 commerce per population, 3 if you're financial

all for no more than 3 upkeep with a courthouse and state property

So in a worst-case scenario, the city will start out with 8 commerce/gold at population 1. If seafood is present it will quickly grow and give much more commerce than it takes in upkeep. Better yet, if you have sufferage or slavery you can rush-build courthouses, libraries, and such to further increase the value. And cultural pressure doesn't matter as long as the city can still work its sea tiles. So the AI isn't nuts to settle these spots. Now defending them can be a bit of a pain if you don't have easy sea access to ship in garrisons. It all goes to show that there's no such thing as a bad food resource. It also goes to show that in the rennaisance/modern eras civ4 actually limits big empires less than civ3 does.

@Zagnut
RCP isn't back in the same way as Civ3. Assuming you don't try to cram everything into the 0-upkeep circle (which isn't a good idea on most maps) you have a good amount of flexibility in placing cities. It's just that there are certain cutoff points that can save you money if you adjust city placement slightly. Also, saving 1 gpt is only a big deal in the beginning when upkeep only comes to 1-2 gpt per city and you economy is rather weak. Once you have 15 cities and are dealing with 6 gpt in city upkeep and have a healthy cash flow, saving 1 gpt probably isn't as valuable as getting a slightly better position anyway. Finally, the new emphasis on resources works against city placement formulas, as the ultimate formula will always be to get as many good resources (especially food) in your city borders as possible. RCP worked so well in civ3 because every tile could yield shields and commerce, unlike civ4.
 
It is an extreme exageration to say we are cursed with RCP again. With RCP your 3rd city had rank 2 instead of 3, your 4th city had rank 2 instead of 4, your 5th city had rank 2 instead of 5 and so on. The benefit for the last city was huge.

The only RCP like thing you can do in civ4 is placing cities just before the distance corruption jumps up from N to N+1; all you gain is 1 gpt per city. You could do this in civ3 too, even after the formula was changed.
 
Requies said:
As I said, my city had that distance maintenance of 1 since it was founded (and it was my second city). And my other city didn't have the distance maintenance even though it's 6 tiles away (farther than every other city). OTOH, all my other cities behaved as you would expect, it's just those cities which were weird....

Req

perhaps roads matter as well in determining distance?

so
N city (founded closer, but with no road connection, city maintenance was 1, by the time it was connected, it grew so city maintenance stayed 1)
E city (founded farther, but roads already in place while it was low population)

Roads could work either by cutting down distance (counting it as moves v. tiles) OR by just saying 'having a trade connection' to the capital.
 
Gato Loco said:
...it seems that the AI still does go for the frozen seafood aisle on the map. Then again, so do I after some consideration. Even a bare water tile can give you 2 commerce when worked. So plop down a fishing town and you can fairly easily get:

+3 commerce from trade routes ( if you have currency + corporation)
+1 gold from shrine income per holy city you own
+3 or +6 commerce equivalent if you run mercantilism
+2 commerce from the city tile itself
+2 commerce per population, 3 if you're financial

all for no more than 3 upkeep with a courthouse and state property

So in a worst-case scenario, the city will start out with 8 commerce/gold at population 1. If seafood is present it will quickly grow and give much more commerce than it takes in upkeep. Better yet, if you have sufferage or slavery you can rush-build courthouses, libraries, and such to further increase the value. And cultural pressure doesn't matter as long as the city can still work its sea tiles. So the AI isn't nuts to settle these spots. Now defending them can be a bit of a pain if you don't have easy sea access to ship in garrisons. It all goes to show that there's no such thing as a bad food resource. It also goes to show that in the rennaisance/modern eras civ4 actually limits big empires less than civ3 does.

With the tiny pockets around another empire, this most frequently happens in the late game, when a bit of extra commerce isn't really going to provide all that much benifit as extra production. In addition to direct maintenance you also have increasing civic upkeep, which likely evens out any marginal bonus to your commerce.

It's most annoying though when the AI plops these down right after you've conquered cities in war, and once the cultural borders expand to normal you assimilate all these villages -- whether you want them or not. Without a way to disband a city by building a settler at 1 or 2 population, there's not much you can do with cultural flips.

Now if you could improve coastal tiles to have some production in addition to commerce, that would be very valuable.

On topic, city maintenance costs are one of the big reasons I utilize Organized for rapid expansion. With half-cost courthouses and lighthouses, you can quickly get pillaged or underdeveloped cities up to speed and cut your maintenance costs dramatically. In addition, it also halves the cost of expensive civics like Police State, Vassalage, and Theocracy. It's interesting that there's a limit to the maintenance added for each additional city...do you know if this is independent of map size?
 
Krikkitone said:
perhaps roads matter as well in determining distance?

so
N city (founded closer, but with no road connection, city maintenance was 1, by the time it was connected, it grew so city maintenance stayed 1)
E city (founded farther, but roads already in place while it was low population)

Roads could work either by cutting down distance (counting it as moves v. tiles) OR by just saying 'having a trade connection' to the capital.

Possible but unlikely. I connected it pretty quickly and I'm pretty sure the maintenance stayed the same. And also why would that city have maintenance while all the rest around the ring (at about the same size) have 0 (e.g., Cumae)?

Req
 
Thalassicus said:
. It's interesting that there's a limit to the maintenance added for each additional city...do you know if this is independent of map size?

Pretty certainly, it is dependent on difficulty level though.
 
Lots'n'lots'n'lots'a great stuff here - but my biggest question concerns map size, which I don't think has been covered in this thread yet (methinks, me may be wrong).

Does Gato or anyone else know what effect the different map sizes have on the OCN or on any of the other results gathered so far?
 
Just to interject a tidbit of some high order strategy, city maintence comes from your generic final gold output.
So this has some interesting effects, the most clear is that the drain isn't any one city but taken as a whole, thus that city to get the iron or some other resource could run at a loss, be largely worthless but still an effective contribution to your empire.
Another one along those lines is that, unlike corruption aka Civ3, this doesn't reduce that cities commerce and thus beakers are unaffected. Meaning that city to get those gems, gold or silver may cost you 6gpt early on in maintence but give you a base 9 beakers in the city with only a few population. You can then overcome the deficit with a holy or gold producing city or other means, all the while having a powerful technology driven empire.


All this does is further strengthen the concept that city placement is driven more by resources than useable land. Of course balancing that with other stuff like this thread's topic is ultimately the "game".
 
I read this thread with great interest - thanks for doing this research and sharing it.

:thanx:

The last few games I have played on Prince, at normal speed, on a variety of small maps (Pangaea, Continents, Ice Age). Mainly so I can complete a game in 3-4 hours.

I tried applying the Cheesy Circle, but found things didn't quite seem to work out the way I expected. So I did a bit of testing in the World Builder (Small Pangaea map). Not very scientific/complete, but here are the results (v1.52).
Spoiler detailed results :

Placing 4 size 1 cities, each 3 spaces away from capital (to N S E & W) was already incurring "distance" maintenance. (Possibly because of map size.)
Total maintenance: 11
Increasing all 5 cities to size 3 -
Total maintenance: 16

Placing 4 size 1 cities 4 spaces away:
Total maintenance: 15
Increasing all 5 cities to size 3:
Total maintenance: 16

Placing 4 size 1 cities 5 spaces away:
Total maintenance: 15
Increasing all 5 cities to size 3:
Total maintenance: 16

The total maintenance includes the civics cost.

Building roads and railways to connect the cities made no difference.

I guess I'll have to do some more testing, would be interested to know if anyone else has looked at the costs on small maps.

Edit:
A little bit more testing, I became curious about the impact of population.
Map was Small Pangaea on Prince.

First test - Capital city only.
Spoiler detail of test 1 :
At size 1, # of cities cost is 0, civics cost is 0.
At size 48, # of cities cost goes to 1, civics cost has reached 10.
At size 112, # of cities cost goes to 2, civics cost has reached 37.
At size 176, # of cities cost goes to 3, civics cost has reached 57.
At size 240, # of cities cost is still 3, civics cost is 82.

Distance cost does not change over this range.

Test 1 Conclusions:
- city size is part of the formula to determine # of cities cost (even when there's only 1 city!)
- total civics cost appears to be a proportion of population, which increases from around 20% with a low population, to around 33% with a high population; the cost is split nearly evenly between the 5 categories of civic (but rounding? appears to add 1 or 2 more cost to the last few sometimes); high upkeep doubles the cost in its category, medium upkeep adds 50% to the cost in its category, no upkeep makes its category free
- obviously these are unrealistic city sizes, but using large numbers may help to show what is going on...

Second test: Two cities, widely spaced (didn't measure distance, but on other end of continent).
Spoiler details of test 2 :
Capital size 1, second city size 239:
Distance cost 8, # or cities cost 6, civics cost 84.

Capital size 239, second city size 1:
Distance cost 0, # of cities cost 7, civics cost 84.

Spreading the 240 population over 3 cities in various combinations made 2 gold difference to the civics cost (observed civics cost range: 84 to 86). Culture, techs and religions seemed to make no difference to any of the costs.

Test 2 Conclusions:
- The civics cost and # of cities costs are both slightly, but not significantly, affected by city placement or the population distribution between cities (but I didn't test with a large number of cities).
- The formula for distance cost IS affected by the population distribution between cities, and these costs are both higher if big population cities are further from the capital.

Third test: Two cities, capital size 239, second city size 1, at various distances.
Spoiler details of test 3 :
At 4 squares from capital (3 empty squares between city and capital), distance cost goes to 1.
At 8 squares from capital, distance cost goes to 2.
At 12 squares from capital, distance cost goes to 3.
At 15 squares from capital, distance cost goes to 4.


Test 3 Conclusions:
- distance affects distance cost (surprise surprise)
- Cheesy Circle is smaller on this map size?/difficulty level?

Fourth test: Increase city size at selected distances.

Spoiler details of test 4 :

Second city 4 squares from capital:
At capital size 1, second city size 1 - distance cost goes to 1
At capital size 1, second city size 8 - distance cost goes to 2
At capital size 1, second city size 16 - distance cost goes to 3

Second city 8 squares from capital:
At capital size 1, second city size 1 - distance is 2
At capital size 1, second city size 5 - distance cost goes to 3
At capital size 1, second city size 8 - distance cost goes to 4
At capital size 1, second city size 12 - distance cost goes to 5

Test 4 Conclusion:
- It seems that the distance cost goes up with size and distance, and that a combination of size and distance makes the cost go up faster. Not surprising really, but it suggests a large empire will get exponentially more expensive, and it also explains why capturing a distant enemy capital early on can be so crippling to one's economy for a while (actually, twice now I have found that early capture of an enemy capital has ruined my economy so much that I could not win the game).

My testing was quite limited, obviously, but I thought I'd post up the results anyway. I didn't go into testing the effect of adding more cities very much ...
 
I've done a bit of testing, and if someone will tell me how to post an excel spreadsheet I can show some of my (incomplete) results. Distance maintenance is affected by distance, population, and dfficulty level. In particular, the "cheesy circle" is smaller on higher difficulty levels and with higher populations. The total maintenance from the number of cities exhibits increasing marginal costs up until the point where the maximum average cost is attained (5 gpt/city at Noble, 8 gpt/city at diety). After that point ("the conqueror's plateau", 24 cities on a standard-sized Noble map, 27 cities on a standard sized diety map), you'll face a constant marginal cost for number of cities maintenance. Number of cities maintenance is not affected by population.
 
StevenJoyce said:
I've done a bit of testing, and if someone will tell me how to post an excel spreadsheet I can show some of my (incomplete) results. Distance maintenance is affected by distance, population, and dfficulty level. In particular, the "cheesy circle" is smaller on higher difficulty levels and with higher populations. The total maintenance from the number of cities exhibits increasing marginal costs up until the point where the maximum average cost is attained (5 gpt/city at Noble, 8 gpt/city at diety). After that point ("the conqueror's plateau", 24 cities on a standard-sized Noble map, 27 cities on a standard sized diety map), you'll face a constant marginal cost for number of cities maintenance. Number of cities maintenance is not affected by population.

You have to zip up excel files as civfanatics doesn't recognize (maybe deliberately) that type of file and then upload the file using manage attachments in the "Advanced" posting post.

Req
 
The problem is when you play above noble level ... I guess AI have some discount in maintance costs. So they can expand more. So I just have to be very small or be agressive to keep up in score. If I'm too small it is just a lot of luck if I get aluminium or not... if not then it is very hard to win space race. Only one component needs copper. But about AI, if human player doesn't expand quickly enough... i mean too slow, then AI just builds spam cities close to my borders and I've no room to expand so I just push culture or declare war.

But now for me culture slider is almost useless, only when you have nothing to research and push for culture victory then it is needed. Now if you run below 100% research you just do something wrong because AI just runs away in tech race. Colosseums and theatres give happiness bonuses but if you play smart enough you can get luxuries or something else to keep people happy. I think it should be as it was in older versions... when you simply couldn't run 100% research in whole game and still were able to keep up in tech. I remember that 20% luxuries was must when you swiched republic/democracy.

I hope firaxis will fix in next patches so 100% science will not be so must than it is now.
 
Desert-Fox said:
I hope firaxis will fix in next patches so 100% science will not be so must than it is now.

Desert-Fox, I think it doesn't matter whether you're on 100% or not.

What matters is how many beakers you're getting (how much science research).

Science slider at 100% of 1,000 earns 1,000 beakers.
Science slider at 60% of 2,000 earns 1,200 beakers.

Guess which one I'd prefer. :) (Ok, and I know it's not quite that simple, but the idea holds true - beakers is what counts, not the %.)

Check the left hand column of your finance advisor (F2) to see how many beakers you're getting. :coffee:

What I find a bit frustrating is how difficult it is to tell what the true cost of that extra city is over time. The city cost in the city screen is garbage for this purpose.

As Gato Loco's tests (and others here) show, the cost of each individual city is disguised by being spread across all cities, and the population component is paid for in Civics. (Not to mention Inflation, has anyone figured out yet how that is calculated - I haven't seen a thread on it yet?)

An example from a recent game - when I founded a new city, I could see that my GPT (gold per turn) went from +4 to -16. So it was obvious that city cost me 20gpt at the time I founded it (ouch). But if I go to the city screen, it says it's costing me only 4gpt!!! So if I only looked at that, the city would appear profitable quite quickly - when it is making 4+gpt - when in fact it needs to be making at least 20! And when the city grows, it will cost more, but I have no way of knowing how much.

All of this makes it very hard to know whether a city is "breaking even" (or was really worth founding), because the cost is hidden (between other cities, civics cost, and possibly inflation too).

StevenJoyce said:
I've done a bit of testing, and if someone will tell me how to post an excel spreadsheet I can show some of my (incomplete) results.

Good work, would love to see the spreadsheet and I'm sure the others following this thread would too.
 
OK, here's what I've found. All tests are preformed on 0-culture cities on a standard-sized, normal speed, Great Plains map. The first sheet deals with city maintenance from the number of cities. It shows the total, marginal, and average cost per city for Noble and Deity levels. The second sheet is a pivot table dealing with distance maintenance. It shows the distance maintenance for a given city as a function of its distance from the capital and its population. This table is pretty incomplete, but I won't have time to finish it for the foreseeable future. The third sheet contains the data underlying the pivot table .

View attachment maintenance-study.zip
 
city upkeep is related to map size.
i just checked WorldSize.xml file and found there're different upkeep discounts on different map sizes.
so i guess it would be easier to analyze city upkeep by modifying map size discount to 100%
 
This thread is the most informative one on city up keep that I have read. Alot of info in there that any civver should use.
 
Yehbo said:
All of this makes it very hard to know whether a city is "breaking even" (or was really worth founding), because the cost is hidden (between other cities, civics cost, and possibly inflation too).

That's the nub of the problem. It's hard to know whether it's economically useful to found a new city or not, without knowing what the marginal cost would be.

There some improvements that could be made to the interface to help out here:

- In the Domestic Advisor screen (or Financial Adviser) include something like:

Next city minimum cost: 12gpt

where the "12gpt" is calculated by comparing your current total income, with what your total income would be if you founded a new city at the closest valid square outside your current borders. In practice it wouldn't be exactly that (you could settle further away than the minimum for a greater distance cost), but it would be a useful yardstick to help decide whether it's worth settling. In this case, I'd need to generate at least 12gpt from worked tiles before this city was profitable, and probably more.

- In the City View screen, include something like:

City profitability: +4gpt

where this number is calculated by comparing your current total income, with what your income would be if the city being viewed was suddenly razed. This would be more relevant information (and include not only city costs but also city income from worked tiles), but you'd only find out *after* settling a city if it was worthwhile. Nonetheless, with enough practice this would be a useful tool to help manage expansion.
 
Top Bottom