cracker
Gil Favor's Sidekick
CAUTION: This post contains Spoiler info for the June GOTM-8 game, so if you have not already discovered the world map and met all the civilizations, then READ NO FURTHER.
.
.
.
.
.
I am not sure if Zouave (or Coracle) ever frequents the GOTM section of things, but I could not resist posting this example of his "Settler Diarrhea" phenomenon.
This screen shot is from the view of Beijing when I encountered the Chinese and set up an embassy with them around 350AD (oops, this said 400BC when I first posted the thread and that was just a braincramp/typo that should have been about 400AD instead of BC) in GOTM-8.
The status of the continent that China is on is almost totally settled solid. There are a few culture gaps that I will try and exploit, but for the most part there are no places for any settlers to be used.
Look at the Garrison for Beijing and you will see three settlers just sitting in the capital waiting for something to do. You will see no offensive military units even though China is the strongest civ on the continent.
After I completed the game play I have been analyzing the saved games and log files to help me better understand how the AI plays the game and it has been very revealing. China had a total of 9 settlers built and sitting in cities with no place to go and these settlers were just tying up resource production potential that could have been more points and more growth per turn. India and Japan were in similar but slightly smaller states of "Settler Constipation".
This is a Monarch level game, so I believe the AI gets some extra military units for free to start the game. The game is also Raging Barbarian activity level.
When the AI founds its first city in this game for each civ, every one of the AI players immediately locks in on producing a settler. What is amazing about this observation is that it is impossible for a player to produce a settler as the first unit from 9 out of 10 start positions. The first city has to grow to a population of 3 citizens in order to produce a settler and on monarch level this growth would require the AI players to produce at least 36 extra food units and 27 shields. The human player would need 40 extra food units and 30 shields.
The conflict comes in the Food vs Shields balance that makes it almost impossible to produce a settler without growing the population first. Most Human strategies, focus on building a unit like a warrior or a spearman, before attempting the first settler.
In the AI cases, they are programmed to fixate on building settlers first, even when it is impossible for them to complete the task.
In this case, the city food and shield production rates per turn were initially at:
Beijing - 2 shields 3 food (bonus grassland with a cow)
Berlin - 2 shields 2 food (bonus grassland and forest)
Delhi - 1 shield 2 food (floodplain)
Kyoto - 3 shields 2 food (grassland with a deer forest)
London - 2 shields 3 food (bonus grassland with a cow)
Moscow - 1 shield 2 food (floodplain)
Paris - 2 shields 2 food (bonus grassland)
All of the start positions were located on rivers with extra food and luxuries close at hand. Because most of the civs would use their workers to build roads and improve terrain, the adjacent squares could improve their contribution beyond the initial values but basically we can look at these starting positions and see that India and Russia need to make it a high priority to increase their shield production rates because they will have enough food to build a settler in 18 or fewer turns, but will only have shields to complete the settler in 27 turns (which is toooooooooo slow). {note that there is the possibility that India and Russia could pop rush a settler to help use up the excess population growth in their capitals due to all the flood plain food and this would help to compensate for the lack of shield producing squares early on.}
Japan has a different problem in that Kyoto is located on grassland and cannot irrigate to increase food production under despotism. This limits growth (until they expand to the cow in turn 8/9) to only 2 food units per turn and makes the earliest expected turn for a settler set at 18 turns. With their high shield production the settler will be "production complete" at turn 9 and then just sit there wasting shields for 9 turns.
What idiot programmed this AI decision tree??? Soren, did you do this?
England and China are in the best positions for this Settler Fixated programming strategy to be effective, because if all else remains constant (in reality shields will actually go up) then the first settlers for these civs could be completed in under 14 turns when the shields are available and anytime after about 12 turns when the food has supported enough population growth.
Paris(AI) and Berlin(human) would both need about 9 or 10 turns to grow the first step and then 6 or 7 turns to grow the next pop point because they gain a wheat and a cow respectively. Their shield production would be available almost at the same time frame.
What is fairly clear from this peek into the AI mind sets, is that producing a settler as the first unit of the build plan is absolutely the wrong decision for over half of these civs.
I should note that the Japanese position may have benefitted from the "join the worker into the city and build settler" gambit.
By fixating on the initial settler build, many of these civs were crippled for the entire rest of the game. Japan wasted over half of their production potential in the early game and ended up in third place on their continent. India was not much ahead of Japan.
On the continent with Germany, England, France and Russia, the Settler fixated initial decision directly caused RUssia to be killed-off in the first 30 turns because the Cathy AI, did not focus on improving production to keep pace with her neighbors. Taking 27 turns to complete your first unit is almost always fatal on Monarch level and above.
.
.
.
.
.
I am not sure if Zouave (or Coracle) ever frequents the GOTM section of things, but I could not resist posting this example of his "Settler Diarrhea" phenomenon.
This screen shot is from the view of Beijing when I encountered the Chinese and set up an embassy with them around 350AD (oops, this said 400BC when I first posted the thread and that was just a braincramp/typo that should have been about 400AD instead of BC) in GOTM-8.
The status of the continent that China is on is almost totally settled solid. There are a few culture gaps that I will try and exploit, but for the most part there are no places for any settlers to be used.
Look at the Garrison for Beijing and you will see three settlers just sitting in the capital waiting for something to do. You will see no offensive military units even though China is the strongest civ on the continent.
After I completed the game play I have been analyzing the saved games and log files to help me better understand how the AI plays the game and it has been very revealing. China had a total of 9 settlers built and sitting in cities with no place to go and these settlers were just tying up resource production potential that could have been more points and more growth per turn. India and Japan were in similar but slightly smaller states of "Settler Constipation".
This is a Monarch level game, so I believe the AI gets some extra military units for free to start the game. The game is also Raging Barbarian activity level.
When the AI founds its first city in this game for each civ, every one of the AI players immediately locks in on producing a settler. What is amazing about this observation is that it is impossible for a player to produce a settler as the first unit from 9 out of 10 start positions. The first city has to grow to a population of 3 citizens in order to produce a settler and on monarch level this growth would require the AI players to produce at least 36 extra food units and 27 shields. The human player would need 40 extra food units and 30 shields.
The conflict comes in the Food vs Shields balance that makes it almost impossible to produce a settler without growing the population first. Most Human strategies, focus on building a unit like a warrior or a spearman, before attempting the first settler.
In the AI cases, they are programmed to fixate on building settlers first, even when it is impossible for them to complete the task.
In this case, the city food and shield production rates per turn were initially at:
Beijing - 2 shields 3 food (bonus grassland with a cow)
Berlin - 2 shields 2 food (bonus grassland and forest)
Delhi - 1 shield 2 food (floodplain)
Kyoto - 3 shields 2 food (grassland with a deer forest)
London - 2 shields 3 food (bonus grassland with a cow)
Moscow - 1 shield 2 food (floodplain)
Paris - 2 shields 2 food (bonus grassland)
All of the start positions were located on rivers with extra food and luxuries close at hand. Because most of the civs would use their workers to build roads and improve terrain, the adjacent squares could improve their contribution beyond the initial values but basically we can look at these starting positions and see that India and Russia need to make it a high priority to increase their shield production rates because they will have enough food to build a settler in 18 or fewer turns, but will only have shields to complete the settler in 27 turns (which is toooooooooo slow). {note that there is the possibility that India and Russia could pop rush a settler to help use up the excess population growth in their capitals due to all the flood plain food and this would help to compensate for the lack of shield producing squares early on.}
Japan has a different problem in that Kyoto is located on grassland and cannot irrigate to increase food production under despotism. This limits growth (until they expand to the cow in turn 8/9) to only 2 food units per turn and makes the earliest expected turn for a settler set at 18 turns. With their high shield production the settler will be "production complete" at turn 9 and then just sit there wasting shields for 9 turns.
What idiot programmed this AI decision tree??? Soren, did you do this?
England and China are in the best positions for this Settler Fixated programming strategy to be effective, because if all else remains constant (in reality shields will actually go up) then the first settlers for these civs could be completed in under 14 turns when the shields are available and anytime after about 12 turns when the food has supported enough population growth.
Paris(AI) and Berlin(human) would both need about 9 or 10 turns to grow the first step and then 6 or 7 turns to grow the next pop point because they gain a wheat and a cow respectively. Their shield production would be available almost at the same time frame.
What is fairly clear from this peek into the AI mind sets, is that producing a settler as the first unit of the build plan is absolutely the wrong decision for over half of these civs.
I should note that the Japanese position may have benefitted from the "join the worker into the city and build settler" gambit.
By fixating on the initial settler build, many of these civs were crippled for the entire rest of the game. Japan wasted over half of their production potential in the early game and ended up in third place on their continent. India was not much ahead of Japan.
On the continent with Germany, England, France and Russia, the Settler fixated initial decision directly caused RUssia to be killed-off in the first 30 turns because the Cathy AI, did not focus on improving production to keep pace with her neighbors. Taking 27 turns to complete your first unit is almost always fatal on Monarch level and above.
