Retried this with every victory condition except Time disabled, because LOL. I am proud to say that I made entirely different mistakes this time! This time the start is a bureaucap, still running Gandhi, and we can rely on there being a god-tier food site somewhere on the map, since it's rainforest.
Normally the number of super specialists one could possibly settle in a city, before winning at least, would be limited by domination, since ideally you would want to bend the entire map to your needs. Disregarding this, there is still the stupid global warming mechanic that quickly converts the map into desert, just like what happened in real life. But what if we disable some victory conditions and dodge the global warming mechanic by not building any forges? Perhaps under these terms, the player is allowed to cover the planet with a thick, muculent paste and just cook specialists for eons.
The cost of a great person may be expressed with sigma notation. For marathon:
C = 300 + Σ(300 * ceiling(n/10))
Where n is 0 → # of GP generated so far. Basically the GPP cost is 300 for the first person and goes up by 300 for persons # 1-10, 600 for 11-20, 900 for 21-30, ... 2400 for 71-80, and so on.
I guess the idea was saturation. My theory was, the map would be able to keep pumping out GPs rapidly even to the end of the game; it would be saturated with cities that had amassed large reserves of points without coughing up a GP. At the end of the game there were, indeed, a bunch of cities that had been barely-beaten to the last GP and were quickly covering the 4200 points of increased incremental cost. Really, though, I believe I corrected some major mistakes from my run in the prior thread but replaced them with new ones.
If I ever did this again I would instead make a guess as to how many great people could be generated, say 150. And make sure approximately
that number of cities was able to reach extremely high population, rather than saturate the map with mid-sized cities. The development here, with 300+ more-uniformly-sized cities, might have made more sense if I went out to 4000 turns rather than 1500.
After teching paper and buying everyone's map, a 20 floodplain site was discovered, occupied by an existing AI.
I donated rice, pig, cow, and deer to Huayna Capac to see if he would grow, saving a settler. Then I blazed a path there and used some heavy horse to apply for ownership of the site. It basically worked out but overall, this session didn't pan out that much better than kaitzilla's game linked earlier.
A general summary of the early game (I sort of tried to stay focused on the workshop tech track):
Iron Working, 2350 BC, expansion may now begin
Currency, 2130 BC, 3 cities
Library pops Great Scientist, 1660 BC
Civil Service, 1240 BC, 8 cities
Oracle Chem, 610 BC, 14 cities
Lib Steam, 30 BC, 24 cities
Communism, 410 AD, 39 cities
Cereal Mills founded, 875 AD, 101 cities
Terrain improvements complete, 1562 AD, 316 cities
Taoism finished spreading, 1612 AD
Golden age initiated turn 702-798 (middle turns)
For some stupid reason I theorized that running a stack of 4 golden ages together, centered around turn 750, would maximize the size of Cuzco. GPs generated this run-
1 academy scientist
1 shrine maiden
1 Mining Inc engineer
1 Cereal Mills merchant
3 scientists, 1 commie spy, 1 music artist, 1 fusion engineer to golden ages
Settled:
130 merchants
4 generals (fascism guy + 3 more)
1 incidental scientist from the National Park city
This comes out to 136 GPs generated by cities, with the 137th requiring 302,400 points.
turn 702/1500 golden age screenie below. There's still a fair amount of growth for even the core cities to accomplish.
Global Warming - The numbers for the global warming computation were as follows.
Number of tiles (Huge map) = 128 × 80 = 10,240
Number of land tiles on this map = 10,213
Number of tiles with forest/jungle = 16
Unhealth from buildings = 1 Forge + 2 Airports = 3
Nuclear events = 0
Constants:
GLOBAL_WARMING_PROB = 20
GLOBAL_WARMING_FOREST = 50
GLOBAL_WARMING_UNHEALTH_WEIGHT = 20
GLOBAL_WARMING_NUKE_WEIGHT = 50
iGlobalWarmingDefense is the forest shield. It is Forests / Land * GLOBAL_WARMING_FOREST = 16 / 10213 * 50
I believe this (~0.078) being an integer comes to zero.
Building unhealth and nuclear history are added together to find the number of times the game "tests" for a global warming event.
Building unhealth is calculated thus: Unhealth * GLOBAL_WARMING_UNHEALTH_WEIGHT / Map Size = 3 * 20 / 10240 ≈ 0.005875
The nuclear event history: Explosions * GLOBAL_WARMING_NUKE_WEIGHT / 100 = 0
The problem with my reading of the code is that numbers may be getting rounded, as integers, during any or all steps of the computation. In any case the value here comes to zero, so no global warming test is performed.
If it were, the chance of global warming in each independent test, would be: GLOBAL_WARMING_PROB - iGlobalWarmingDefense / 100
... Which is 20% per test for my defenseless map. To get from 20 to 19 percent there would need to be 205 forests, and to negate all sources of global warming there would need to be 4086 forests, or more generally, 40 percent coverage.
TL;DR
If my number crunching is correct, a nuke-free, Huge-sized map can absorb 255
from buildings, and I went totally overboard avoiding forges. But I'm skeptical about how much buildings in civ4 really help you, given their cost, and I don't think I lost many— if any— turns in the key development phase by avoiding them. I eventually had to build 1 forge as a prereq for the SoL. But optimally I should have put industrial parks in select cities, after all else was accomplished, to boost GP farming while remaining under this 255
benchmark.
Here's some words about the rainforest script. This is a flat map script by default that generates a massive grassland (green) covered with jungle. Other features include blobs of mountains surrounded by hills, blobs of desert surrounded by plains, and a couple monstrous rivers that zigzag acrosss the map. Banana, Rice, and Pig are the dominant resource clusters on this script. It tends to be metal-poor and quite ridiculously food-rich. The sea level setting affects the map even though it is greyed-out. The climate setting does not.
Example rainforest food resource packing
Archipelago has far more relevant resources to a relatively weaker corp (for our purposes). I wonder if the sushi corp on archipelago can beat the cereal mills + massive farmland on rainforest? Sushi gives less food per resource, and water tiles are inferior to the land you get in rainforest. However, there is far more seafood in total in Archipelago than there are grains in Rainforest.
Turn-by-turn anim: AIs struggle against the jungle, get brushed aside with armor, and 400 Fast Workers sweep in bringing peace, harmony, and trusted brand name products to the area.
I don't mind improving tiles, that part of the game was kind of fun. Do you know what was an absolute Bee Eye Itch? Spreading Taoism and the Corporations. I built 3 monasteries, and two cities spammed mishes for the entire first 600 turns. I bought them airports to help. I knew that dealing with the corps and having to zigzag universal suffrage would be hell, and it was.
Gandhi's SPI trait pulled freight in this game. In marathon you can basically run slavery and caste system at the same time; later rep and suffrage.
National Park site below. The park and the epic were bought directly after the Kremlin was built. The epic may have been a mistake, hard to say exactly.
I built wall street (below) but not oxford. Oxford was useless in this campaign.