Page 1 of 1

Unbalanced Geographical Distribution of Core Civs

Posted: Mon Jul 16, 2018 11:36 pm
by vodot
Hi all, here is a rough approximation of our current geographical distribution of Core civs:
civloc.PNG
I think we need to lose a couple European civs and/or add a few Pacific & African civs. On an Earth map with fixed starting positions, this creates an Australian promised land and a massive, mostly-empty Africa, Asia, South America, and Canada. I need to mess around more with the EARTH scenarios/maps out there more to see how most people are addressing this, but it seems like the following additions to the 'Core' would be good ones from a Geographical and prominence perspective:

In Oceania:
-Australian
-Aborginies
-Indonesian

In Africa:
-Congolese
-Maasai
-Morrocan

In South America:
- Chimu

In Northern(ish) Asia:
- Scythian

In India:
- Vedic

In China:
- Xiongnu

In North America:
- Inuit
- Aluet
- Canadian

Re: Unbalanced Geographical Distribution of Core Civs

Posted: Tue Jul 17, 2018 12:28 am
by JTN
Some thoughts:

The original selection of the Core nation group is basically inherited from the original Civ games, so inherits their biases. See gna patch #3432 for rationale. (It's not much of an excuse, but it's an excuse.)

Changing the Core set is a bit fraught (noted in patch #3448). We can basically never remove a nation from Core once we've added it, for savegame compatibility reasons.

We can in principle promote new ones to Core, if we are sure we won't regret it. That mainly puts extra burden on translators. I don't have a feeling for how much translators are actually using the split between Core and Extended to focus their work (the translator whose comments prompted this whole split has long since wandered off), and how much players using those translations are benefiting.

It seems a bit odd to optimise the core set (affecting multiple rulesets forever) around what is essentially one scenario out of many. Another solution would be to get the engine to choose an evenly distributed set of start positions, as vaguely mooted in hrm #690313; so someone would always get Australia. This would benefit other scenarios, such as Rhue's Europe with ~200 starting positions (which are as unevenly distributed as you might imagine).

...have you got the lat/long data you used for that map? I've had a vague notion of a spatial+temporal nation picker in the back of my mind for years, although it is way down the list of sensible priorities.

Re: Unbalanced Geographical Distribution of Core Civs

Posted: Tue Jul 17, 2018 6:11 pm
by vodot
JTN wrote:Some thoughts:

The original selection of the Core nation group is basically inherited from the original Civ games, so inherits their biases. See gna patch #3432 for rationale. (It's not much of an excuse, but it's an excuse.)
haha I sorta figured.
JTN wrote:Changing the Core set is a bit fraught (noted in patch #3448). We can basically never remove a nation from Core once we've added it, for savegame compatibility reasons.

We can in principle promote new ones to Core, if we are sure we won't regret it. That mainly puts extra burden on translators. I don't have a feeling for how much translators are actually using the split between Core and Extended to focus their work (the translator whose comments prompted this whole split has long since wandered off), and how much players using those translations are benefiting.
Right. Hm. I would have wanted to remove some European civs to offset adding new civs. I didn't realize that we couldn't remove civs from core without breaking backwards compatibility. Could we deprecate support for a subset of extraneous European civs over the course of a few releases? Target 3.0 for a revised core set, or something?

Demote --> Promote:
  • Austrians --> Aboriginies
  • Frisians --> Australians
  • Byzantines (or polish? sorry poles...) --> Scythian
  • Danes --> Vedic (Northern India)
  • Cherokee --> Inuit
JTN wrote:It seems a bit odd to optimise the core set (affecting multiple rulesets forever) around what is essentially one scenario out of many.
Yeah, but EARTH is really more of a super-scenario, isn't it? It deserves special treatment, I think (ignoring whether or not we have any bandwidth to acutally give it that treatment...
JTN wrote:Another solution would be to get the engine to choose an evenly distributed set of start positions, as vaguely mooted in hrm #690313; so someone would always get Australia. This would benefit other scenarios, such as Rhue's Europe with ~200 starting positions (which are as unevenly distributed as you might imagine).

...have you got the lat/long data you used for that map? I've had a vague notion of a spatial+temporal nation picker in the back of my mind for years, although it is way down the list of sensible priorities.
Yeah, that sounds pretty awesome; I attached the data but it's just an XY plot, and it's a guess. I suppose I could convert it to lat long, but XY as a ratio of map XY is probably better, right? Then we could round it off to an integer and maybe auto-find the nearest land tile?

Civ,x,y
ethiopian,58,40
carthaginian,45,65
egyptian,50,65
mali,45,50
songhai,40,50
zulu,50,20
mongol,75,80
chinese,83,60
korean,85,70
japanese,89,70
russian,53,85
indian,70,55
thai,78,40
khmer,78,45
sumerian,55,65
babylonian,50,65
arab,55,60
persian,60,65
hittite,60,70
turk,55,70
byzantium,50,78
hunnic,50,85
austrian,48,80
french,44,80
celtic,49,83
polish,50,86
frisian,46,83
german,47,84
dutch,45,85
danish,45,87
viking,45,90
swedish,45,88
english,41,85
italian,45,77
roman,46,75
greek,49,73
portuguese,40,73
spanish,42,74
cherokee,3,65
sioux,3,73
american,15,70
iroquoi,11,70
mayan,11,45
aztec,7,55
taino,17,55
apache,5,75
polynesian,95,28
tupi,25,25
brazilian,27,27
inca,15,37