Please look at the results of a town in your math -- it has negative cost food. @_@
Or take a look at a size 1 financial city with a lighthouse and the colossis -- 2 food,
0 cost. That one citizen can produce the full food yield of the nearby water at zero GDP cost...
Ie, your math is screwed.
![Smile :) :)](/data/assets/smilies/smile.gif)
That citizen shouldn't be able to fish the entire surrounding ocean all by himself.
Note that if you changed the GDP per citizen to 10, this singularity doesn't happen. But you still get other strange effects -- a town producing 7 coin 2 food 1 hammer only costs 3 coin per square.
So 3 coin produces (2 food 1 hammer).
So a 6 size city could work
20 towns, generating 40 food and 20 hammers.
12 of that gets eaten, leaving 28 surplus food. From a size 6 city.
A size 6 city under with lots of towns could produce 2 surplus food, 7 hammers, and 43 coin. This is a very different place than your model.
Note that your hammer system is also quite broken -- having an extra hammer source can quite easily reduce the productivity of a city, even if the city doesn't want to use the tile.
--
1: The basic unit of production in civ4 WORK not COIN. Coin is a mechanism of exchange, research and wealth.
As a real-world example, the Spanish empire fell apart from having too much Coin back home, and not enough work.
2: Working a tile produces it's output -- this takes effort on the part of the citizens, and imported coin from the Empire isn't enough.
Now, working tiles "fractionally" makes sense.
3: Commerce is not GDP. Commerce represents wealth, luxuries, trade and the ability to engage in research and culture.
--
Let's say a city can produce 10 units of work per citizen. We'll leave the base (2 food + 1 shield + 1 commerce) amount alone.
1 unit of work can thus provide:
1/10th of a specialist, if we have them.
1/10th of the yield of any square in the domain of the city.
Each of the specialist and squares provides a tradeoff of work for resources.
This generates a reasonably complex set of vectors in a 6 dimensional space representing work
![Frown :( :(](/data/assets/smilies/frown.gif)
food, production, coin, research, culture) tradeoffs. (Note that raw commerce can be easily converted into (coin, research, culture) given your empire and city settings)
A 4 food tile is 0.4 food per unit of work.
A 1 food 1 hammer 4 commerce tile is (0.1 food, 0.1 hammer, 0.4 commerce) per unit of work.
The delta tradeoff is (0.3 food) for (0.1 hammer 0.4 commerce).
Add in a mine at (4 hammers) gives us (0.4 food) for (0.4 hammers) -- or, locally, a 1 food for 1 hammer tradeoff.
Plug that back into the previous case, and we get (0.2 food) for (0.4 commerce) -- or 1 food for 2 commerce tradeoff.
Given even a handful of slightly different tiles, these cost tradeoffs fall out. Those with the best tradeoffs win.
More importantly, if you set "how much do I value food, research, culture, coin and hammers" at a given location, it is easy for the AI to greedily grab the best tiles or specialists for that location that maximize the result. A value-curve is harder to solve, but can be approximated by good old binary search/newton's method.
One would set a "global" value for each of those things, and allow local fudge-factors. Add in the ability for cities to import goods automatically (if one city is willing to pay enough food to pay for the transport costs, it automatically starts importing) and we would have an economy. You could even choose to "favour" a city with imperial taxes and let it send it's money to other cities in exchange.
Note that moving goods around has a cost: in an ancient era, moving food from one area to another would generate significant spoilage, and in the modern era it requires lots of infrastructure (coin and/or hammers).
Coin would be the easiest good to transport long distances by far -- so paying for things in coin would be the natural thing to do. Especially if you tweaked the economy of the game so that hitting 100% research was much harder (or allow negative coin rates).
...
In either case, this generates a rather cool AI improvement for the current game.
Very good idea, dispite my criticisms!