Analysis of commerce for Towns vs Coast

[Zombie69]

Just to be clear, i agree that in your example of a city with only one sea food resource and no useable land tiles whatsoever, it's best to make the lighthouse first. I just don't agree that you should want to have such a city at all! Therefore, in all cities that i would want to have, a granary should always be made first.

 

[DaviddesJ]

Remember that i was talking about a city with 2 sea food resources.

No, I just recall what you actually wrote, which was, "In Civ 4, i would never, ever slave rush anything without a granary, except for slave rushing the granary itself!" The words "never, ever" seem to apply more broadly than "except when there are fewer than 2 food resources"!

I agree that towns with 2 sea resources have so much food that a quick granary gives the highest payoff. But such towns are rare, at least on standard maps.

Just to be clear, i agree that in your example of a city with only one sea food resource and no useable land tiles whatsoever, it's best to make the lighthouse first. I just don't agree that you should want to have such a city at all! Therefore, in all cities that i would want to have, a granary should always be made first.

OK. I guess this is internally consistent, but it seems very strange to me. Such a town might easily generate 20-30 commerce, way more than it costs to support. Why not build it?

 

[Zombie69]

Read my previous posts in this very thread. The quote above was from a post expanding on the concept of a city with 2 sea food resources. Besides, as you can see from my last post, the concept you presented is indeed a case where lighthouses should be made before granaries; however, i would not want such a city in my empire, and therefore my general statement stands, and i'll always rush the granary first.

If you want to found such cities, then yes, make the lighthouse first. But i won't found those cities and so in my games, i will never rush anything before the granary!

Note that simply having 2 or 3 unimproved grassland tiles would make the granary worth rushing first.

 

[DaviddesJ]

Note that simply having 2 or 3 unimproved grassland tiles would make the granary worth rushing first.

This is not clear. The advantage of the lighthouse is that you can work coast instead of grassland for extra commerce (but no extra food). Whether this is better or worse than the granary depends on the relative value of commerce now vs long-term growth.

 

[Zombie69]

Remember that either way, you'll want to have both the granary and the lighthouse as soon as possible, rushing both. The only question is which to rush first. If i only have one sea food resource, then i'll have no complex using plain grassland until after both are built, which with slavery only amounts to a few turns. This means a little less commerce for a few turns, but faster growth, having both structures set up sooner, starting on the library sooner, etc. Because of extra pop, you get more commerce with this method before long, so it doesn't take many turns before you've surpassed the other method in total commerce (and you still have more pop to show for it).

I'd have to be hurting really bad for commerce NOW to want to use coast rather than grassland while pop rushing a structure that will let me have the advantage of both tiles in one (i.e. 2 food + commerce).

Without slavery, yes i might choose coastal tiles at 1 food over grassland. But with slavery, the extra food is too good to pass up.

 

[alexti2]

If the AI has better cities to trade with, you'll get better trade routes, and the AI typically will have better cities as the game progresses.

Usually AI has little problems having good cities early on. It's usually your own city that limits the route income.

 

[alexti2]

Just to be clear, i agree that in your example of a city with only one sea food resource and no useable land tiles whatsoever, it's best to make the lighthouse first. I just don't agree that you should want to have such a city at all!

That's not very practical. If you just sit in your capital and refuse to build any cities because the available land (or rather sea) doesn't meet your strict criteria, you're not likely to win on harder difficulties. Even poor 1-tile island with one seafood nearby can host a city that will generate some research and gold for you. Actually, it's not even that bad for that purpose.

 

[Zombie69]

True. However, there are normally better spots to take, or better things to do than to make settlers to go there. I'm not saying that those cities don't pay for themselves. It's just that they take longer to do so, and i'll always have better options available. So in my games, i've never had to take those spots.

I have yet to play an island type map. Maybe there aren't as many good spots in those maps and i'll have to take some of those. I'll try such a game soon and we'll see.

 

[Wodan]

As for bigger is better, is still is, and very much so. No matter how many cities i have, adding an extra city on a one-tile island with 2 sea food ressources will always pay off. The extra commerce gained from the sea far outweighs the added maintenance cost. Slave rush a granary, a lighthouse, a library and a university (in that order) and the city will be so good that you'll want to add an academy to it the next time you get a Great Scientist. Especially if you're playing a financial civ.

I think this overlooks several things:
-- cost of building the Settler (say you instead leave that city on generate research or gold; that's an extra 500 beakers or so you could have... how long before your new city will have generated 500 beakers?)
-- increased maintenance in ALL your cities
-- cost of building the garrison units (again, you could have been generating research)
-- cost of unit maintenance for garrison

Oh, sure, at some point ANY city is going to become a net positive gain. The question is whether that point will come before the end of the game.

Wodan

 

[Quantum7]

I think this overlooks several things:
-- cost of building the Settler (say you instead leave that city on generate research or gold; that's an extra 500 beakers or so you could have... how long before your new city will have generated 500 beakers?)
-- increased maintenance in ALL your cities
-- cost of building the garrison units (again, you could have been generating research)
-- cost of unit maintenance for garrison

Oh, sure, at some point ANY city is going to become a net positive gain. The question is whether that point will come before the end of the game.

Wodan

- The cost of your 1st and 2nd point shouldn't be that extreme. Build the units in a production city or chop it, etcetera. --> Both together shouldn't cost more than 200 science.
- 3rd point is something you should take into account when considering whether the city will be worthwhile.
- 4th point is interesting, but probably not relevant. Isn't the number of units you can maintain dependent on the number of cities you have? In that case it would even be beneficial to have an extra city.

With regards to your question: You're absolutely right. Only the moment you start getting pure gain from it is probably less than 50 turns or so. Might be interesting to do calculations on that .

 

[alexti2]

I think this overlooks several things:
-- cost of building the Settler (say you instead leave that city on generate research or gold; that's an extra 500 beakers or so you could have... how long before your new city will have generated 500 beakers?)
-- increased maintenance in ALL your cities
-- cost of building the garrison units (again, you could have been generating research)
-- cost of unit maintenance for garrison

I assume that you meant 50 beakers instead of a settler? If we look at what the new city will bring in the assumptions of the evaluation, we will initially get 1C of original tile + 2 commerce from coast + 4 trade routes*1 (assuming 2 natural routes +2 from GL). That's 7 commerce per turn. Against that you'll have to pay more maintenance. Because of that the result will depend on map size and how many cities you already have. For the first few cities, they'll pay off within 10 turns or so, but as you build more, the time to recover the costs will be getting longer, so at certain point it becomes better not to build new cities (in practice at that stage, it's usually better to take over AI cities which start with decent population and some buildings)

In this comparison, we can assume the cost of building garrision unit to be 0, because it's done from the production that wouldn't exist otherwise.

I believe that the more cities you have, the more free units you get (it might depend on the civics though), so adding a new cities will often have positive effect on unit maintenance.

 

[Wodan]

500, 50, whichever. I just pulled that number out of the air. I suppose I was thinking an "average" city with both production and tech. If you had that city set on "Research", in the 4-5 turns it would take to pump out the Settler, it definitely would be in the hundreds (of lost Beakers).

"garrison cost = 0, becaue done from production that wouldn't exist otherwise". I don't get this, Alexti. Any new city is busy building infrastructure (Granary, Library, etc.) No way you would build your garrison there.

Good points from everyone. Another city would negate unit maint.

Wodan

 

[alexti2]

500, 50, whichever. I just pulled that number out of the air. I suppose I was thinking an "average" city with both production and tech. If you had that city set on "Research", in the 4-5 turns it would take to pump out the Settler, it definitely would be in the hundreds (of lost Beakers).

The city doesn't stop researching when you build settler. It only stops to grow. If instead of building settler you set your city to build research (1 shield=1/2 beaker), you'll get around 50 beakers instead of a settler (some factor also influense it, so typically it will be somewhere in 20-80 range.

"garrison cost = 0, becaue done from production that wouldn't exist otherwise". I don't get this, Alexti. Any new city is busy building infrastructure (Granary, Library, etc.) No way you would build your garrison there.

You can build garrison there - that guarantees you garrison cost of zero. If you prefer to build something else (which you often do) instead of garrison, that's presumably because you expect more benefits from that, thus you're even extracting greater value. Perhaps, the city will eventually build some units which will become a garrison for a newer city. The key here is that you've an option to build garrison locally, which guarantees that the cost of garrison in this strategy will be no more than zero. It sounds weird to have negative garrison cost, but it means that you may have an option to exchange your zero-cost garrison to something more valueable.

 

[Quantum7]

The city doesn't stop researching when you build settler. It only stops to grow. If instead of building settler you set your city to build research (1 shield=1/2 beaker), you'll get around 50 beakers instead of a settler (some factor also influense it, so typically it will be somewhere in 20-80 range.

I actually assumed he was referring to using other tiles for producing the settler than for producing techs. In example extra hill tiles instead of cottages . --> Bigger effect.

 

[alexti2]

I actually assumed he was referring to using other tiles for producing the settler than for producing techs. In example extra hill tiles instead of cottages . --> Bigger effect.

True, makes it hard to calculate the difference though. What might be an average number of extra commerce tiles and what would be their income? Let's consider the early phases, where the most of expansion happens. If the settler is built by production city (for food city the calculations are very similar, and there's no point in building settler in commercial city), let's assume the city is setup to 1 food source (5 units) and 3 mined grassland hills. So it's 10 production + 2 extra food, so the settler production will take 8.33 turns. What can be reasonably expected to be more commerce efficient? Unused worked cottages are unlikely, so perhaps it will be the coast tiles. So we can switch 3 hills to 3 coast tiles. That will give +6 commerce per turn. If we have library here it will give us 8 beakers. We may not have library in production city, but we may have monastery, or we might be financial. So 8 beakers looks somewhat reasonable estimate. That will give us 67 beakers instead of settler.

If we look at a larger production city, for example with 2 food source and 6 mined grassland hills, it will have production of 23 (assuming forge) + 2 extra food. So it's 4 turns to build a settler. Instead we can get 12 commerce per turn. We assume that by that stage we have both library and university. So it will be 18 beakers per turn - to a total of 72 beakers instead of settler.

So it doesn't look like you can squeeze much more research using this method.

 

[DynamicSpirit]

(This was about building cities on tiny islands)

True. However, there are normally better spots to take, or better things to do than to make settlers to go there. I'm not saying that those cities don't pay for themselves. It's just that they take longer to do so, and i'll always have better options available. So in my games, i've never had to take those spots.

Slightly changing subject, but two other reasons why I'm very reluctant to build cities on islands are:

1. Most islands are places I find when my ships are finally exploring the oceans. They are nowhere near my capital and will have far higher distance costs than a lot of spare spots on my home continent. (I learned this the hard way a while back when I was still in Civ3-expand-everywhere-mindset and a city I found a couple of islands, immediately got settler there 'before the AI does it'. City started with distance cost 9, no. of cities cost 2 - total cost of 11, even before you add in the increased civics etc. cost)

2. Much harder to defend. If it really is a single square in an ocean then you are forcing the AI to attack from the sea which gives it a disadvantage, but there's still a serious issue with numbers of units: On a continent I can get away with a smallish ratio of no. of military units/no. of cities because I know that if the AI attacks I can (with good roads/railways) very quickly get lots more units to the battlefield from nearby cities, and if the city is inland and has good cultural borders I'm also likely to get 2-3 turns of war to prepare before the enemy forces can reach the city anyway. Plus if necessary there are also a lot of cities that can rushbuild more units to get to the battle in 2-3 turns. On an island you can't - if you want to keep the city you have to keep however many units it'll take to defend that city *permanently* stationed there. The AI can suddenly declare war, its galleons or transports can sail up and unload its units next to your city all in one turn, and, other than rush-building one extra unit/turn till you run out of gold/pop, you've no means of getting any extra units there for at least 4-5 turns, depending how far the island is from your mainland.

 

[Zombie69]

This quote was about building cities on tiny islands with less than 2 sea food tiles, not all tiny islands in general. With 2 or more sea food tiles, i don't think twice and build there before the AI can beat me to it. Of course, this assumes the distance is reasonable, but the same applies to any city, not just islands.