This part of the system has many bugs.
I am currently testing various workarounds to those errors, but it seems that I only find even more problems.
1. A reference for an already reported bug.
Appeal does not matter when AI evaluates Earth Goddess
2. Fundamental design issue - AI scores only the current situation.
No thought is given for the future. Which is wrong because AI will never choose something that gives benefits in the future, even very near. Pantheons like Divine Spark are scored as 0 until a District is built (which is often too late), God of Healing and God of War - always 0, etc.
Example.
Pantheon God of the Open Sky. Gives +1 Culture from pastures.
Let's say there are 5 Cattles within city borders. Crazy good situation. AI will score this as 0 (zero) because these are not Pastures yet. If they build 1 pasture, it will be scored as 46 (1 yield), the potential 4 other tiles will not be scored.
Virtually every pantheon that is based on Improvements (there are quite few) is scored as 0 as long as an actual Improvement is built. No matter how many resources are around.
3. Only the recently settled city is scored.
Yup, I would think that if AI settles 2nd city, at least this scoring of the current situation will evaluate properly benefits from 2 cities. Nope. When the 2nd city is placed, scoring starts showing only benefits from 2nd city, forgetting about Capital (which is much more developed already, has more tiles, etc.) AI essentialy hurts itself.
4. Many modifier types and requirements are ignored when doing a scoring. Ignored or, in case of reqs, they default to always being true. Examples.
REQUIREMENT_GAME_ERA_IS - ignored, always true, pantheon scores even effects from different eras
REQUIREMENT_PLOT_IS_LAKE - lakes are not recognized, basically all city tiles are considered a lake
REQUIREMENT_PLOT_TERRAIN_CLASS_MATCHES - ignored, all tiles are scored
REQUIREMENT_PLOT_HAS_ANY_FEATURE - not working
REQUIREMENT_PLOT_FEATURE_TAG_MATCHES - not working
REQUIREMENT_PLOT_RESOURCE_VISIBLE - not working, always true
MODIFIER_ALL_CITIES_ADJUST_CITY_YIELD_PER_POPULATION - generally any modifier that grants yield per population is broken as it doesn't calculate properly non-integers. "0.2" is rounded down to "0", so no matter how many pop the city has - always is scored as 0. Only when it is "1.0", it is scored properly.
MODIFIER_SINGLE_CITY_ADJUST_BUILDING_PRODUCTION_MODIFIER - always scored as 0, even if a city is building something; which is certailny a bug because a similar modifier used by Monument of the Gods to grant production for Wonders, is working ok. When a city build a Wonder, it scores >0. [Edit. It seems that Wonders are also scored as 0. See later]
5. Some Pantheons have unjustified huge scoring.
This is a problem "in the other direction". The rule seems very simple - usually 1 yield = 46 points. I tested it with current pantheons, any many of my own and it holds.
However, the Pantheons for adjacency like "+1 faith to Holy Site for adjacent Desert tile" or "tundra tile", etc. produce scoring that can reach more than 1000 points. The biggest I've seen was around 1500 points. It equals to ~30 Faith per turn.
I don't know what the issue is here, but I suspect that the scoring checks this time all possible tiles and sums up all possible adjacencies. So, if there are 3-4 desert tiles, HS can be placed in many tiles on and around and in total all adjacencies will give very high yield, reaching easily several. Which is of course wrong because there can be only 1 holy site per city, and adjacencies like 2-3 should produce scoring like 92-138. The biggest one, with adjcency 6 (HS in the middle of a big jungle or desert) should give max. 276. Any bigger result is just wrong.
I am currently testing various workarounds to those errors, but it seems that I only find even more problems.
1. A reference for an already reported bug.
Appeal does not matter when AI evaluates Earth Goddess
2. Fundamental design issue - AI scores only the current situation.
No thought is given for the future. Which is wrong because AI will never choose something that gives benefits in the future, even very near. Pantheons like Divine Spark are scored as 0 until a District is built (which is often too late), God of Healing and God of War - always 0, etc.
Example.
Pantheon God of the Open Sky. Gives +1 Culture from pastures.
Let's say there are 5 Cattles within city borders. Crazy good situation. AI will score this as 0 (zero) because these are not Pastures yet. If they build 1 pasture, it will be scored as 46 (1 yield), the potential 4 other tiles will not be scored.
Virtually every pantheon that is based on Improvements (there are quite few) is scored as 0 as long as an actual Improvement is built. No matter how many resources are around.
3. Only the recently settled city is scored.
Yup, I would think that if AI settles 2nd city, at least this scoring of the current situation will evaluate properly benefits from 2 cities. Nope. When the 2nd city is placed, scoring starts showing only benefits from 2nd city, forgetting about Capital (which is much more developed already, has more tiles, etc.) AI essentialy hurts itself.
4. Many modifier types and requirements are ignored when doing a scoring. Ignored or, in case of reqs, they default to always being true. Examples.
REQUIREMENT_GAME_ERA_IS - ignored, always true, pantheon scores even effects from different eras
REQUIREMENT_PLOT_IS_LAKE - lakes are not recognized, basically all city tiles are considered a lake
REQUIREMENT_PLOT_TERRAIN_CLASS_MATCHES - ignored, all tiles are scored
REQUIREMENT_PLOT_HAS_ANY_FEATURE - not working
REQUIREMENT_PLOT_FEATURE_TAG_MATCHES - not working
REQUIREMENT_PLOT_RESOURCE_VISIBLE - not working, always true
MODIFIER_ALL_CITIES_ADJUST_CITY_YIELD_PER_POPULATION - generally any modifier that grants yield per population is broken as it doesn't calculate properly non-integers. "0.2" is rounded down to "0", so no matter how many pop the city has - always is scored as 0. Only when it is "1.0", it is scored properly.
MODIFIER_SINGLE_CITY_ADJUST_BUILDING_PRODUCTION_MODIFIER - always scored as 0, even if a city is building something; which is certailny a bug because a similar modifier used by Monument of the Gods to grant production for Wonders, is working ok. When a city build a Wonder, it scores >0. [Edit. It seems that Wonders are also scored as 0. See later]
5. Some Pantheons have unjustified huge scoring.
This is a problem "in the other direction". The rule seems very simple - usually 1 yield = 46 points. I tested it with current pantheons, any many of my own and it holds.
However, the Pantheons for adjacency like "+1 faith to Holy Site for adjacent Desert tile" or "tundra tile", etc. produce scoring that can reach more than 1000 points. The biggest I've seen was around 1500 points. It equals to ~30 Faith per turn.
I don't know what the issue is here, but I suspect that the scoring checks this time all possible tiles and sums up all possible adjacencies. So, if there are 3-4 desert tiles, HS can be placed in many tiles on and around and in total all adjacencies will give very high yield, reaching easily several. Which is of course wrong because there can be only 1 holy site per city, and adjacencies like 2-3 should produce scoring like 92-138. The biggest one, with adjcency 6 (HS in the middle of a big jungle or desert) should give max. 276. Any bigger result is just wrong.
Last edited:
