Poor starting positions on new map

[Snofru1]

While testing my newest space map (Large 07) I found out that many civ starting positions are very poor. Even though there is still a lot of open space on the big continents the game seems to favour small islands, e.g. isolated arctic islands with only space for 2 cities. Once I even had a starting position on a 1 tile island (there is a civ on a 2 tile island in the attached savegame).

@Toffer90 , maybe you can check if you can improve on this.

You can easily see where the starting positions are if you start the attached savegame (SVN 11390) and open worldbuilder.

 

[Toffer90]

Noted, I might look into it around the next weekend.

Edit: I take it this is with the world mapscript, since you didn't specify.

 

[Snofru1]

Noted, I might look into it around the next weekend.

Edit: I take it this is with the world mapscript, since you didn't specify.

This was on a space map scenario but the "earth" part of it was made with the "World" mapscript. I will check if see similar starting positions on a new map in a custom game and come back...

 

[Snofru1]

OK starting positions are ok on a new custom game with the "World" mapscript.
Starting positions are also ok when I make a scenario with non-fixed starting positions out of this map.

But the starting positions (not all but many) are poor when I make a space map scenario out of the same map (see the savegame in the 1st posting). Using the "No Longitude and Langitude Limits" modmod doesn't change this.

 

[Toffer90]

OK starting positions are ok on a new custom game with the "World" mapscript.
Starting positions are also ok when I make a scenario with non-fixed starting positions out of this map.

But the starting positions (not all but many) are poor when I make a space map scenario out of the same map (see the savegame in the 1st posting). Using the "No Longitude and Langitude Limits" modmod doesn't change this.

Right, I assume you don't use the starting position world mapscript established when you cut it into your space map scenario file, so the generic starting position code in dll decides where the "alive" players should start when starting a game on the scenario.
Not sure I want to work on the generic dll implementation for starting plot assignment too much.

 

[Snofru1]

Right, I assume you don't use the starting position world mapscript established when you cut it into your space map scenario file, so the generic starting position code in dll decides where the "alive" players should start when starting a game on the scenario.

Right. I started with doing so when new civs didn't start somewhere in space anymore after some fix made in the past.
Surprising for me is, that when I create a map with "World", save it in worldbuilder, remove all the units and starting positions and save it as a scenario I don't have this issue. Here I also use the generic code in dll but the positions are fine. When I insert this identical map in my space map template the dll starting positions are in the bad places (at least not in space anymore!).

Not sure I want to work on the generic dll implementation for starting plot assignment too much.

I can understand this totally. If you won't work on this I think I have to go back changing the space maps to the starting positions created by the "world" mapscript.

 

[Toffer90]

Right. I started with doing so when new civs didn't start somewhere in space anymore after some fix made in the past.
Surprising for me is, that when I create a map with "World", save it in worldbuilder, remove all the units and starting positions and save it as a scenario I don't have this issue. Here I also use the generic code in dll but the positions are fine. When I insert this identical map in my space map template the dll starting positions are in the bad places (at least not in space anymore!).

Maybe its attempt to spread starting positions far apart from each other gets confused by the huge map size, so it overshoots in some calculations by using the length of the longest plot distance axis (south to north) as a variable in the calculation for expected average distance between starting plots. That way it really goes out of its way to spread civs away from each other so that landmass value becomes a less significant factor compared to the value of being X plots from neighbours. value of starting positions are normalized over an ideal distance to neighbour factor that is. So low value starting plots gets added value for being far away from the high value starting plots, while high value starting plots are reduced if they are close to other high value starting plots.

 

[Snofru1]

Another thing I just found out is that with smaller number of civs (10 for the large map) all starting positions look fine. But if you add more (tested up to 21) they become worse. That probably fits quite well to your theory, Toffer.

 

[tramwajg]

could be interesting feature of mod if there were more earthlike planets with possible starts for civilizations (or even with different ecosystems and vegetation) although I can imagine it could be hard on available map sizes in what part of space were these civilizations starting? sounds interesting, I remember old maps descriptions where was written "don't use more than 10 civs or some civs will start in weird spots"

I can say after playing on your continents maps many times with >20 AI that my start was always good with a lot of land available, although might be just by chance

 

[Snofru1]

The first sets of the duel space maps had a kind of "2nd earth" in space but it created a lot of trouble. The "don't use more than 10 civs" thing was when civs could still spawn in space but that luckily was repaired a while ago.

Good to hear that your starts were good but there surely must have been AI sitting like lame ducks on some islands. In one game my own starting spot (21 starting civs) I sat on a one tile island - that would have been a challenge !
And I also remember your quite bad starting on the Archipelago map, though I have seen worse ones, too.
I really have to consider what to do about this. Going back to pre-fixed starting locations means that the player will always start on the same tile iirc, not too great a solution...

 

[tramwajg]

https://forums.civfanatics.com/threads/space-pangea-nightmare-eternity-upscaled-research-d.672755/ it was my gameplay on your large pangaea, https://i.ibb.co/wBZD4ZR/turn-2178.png one AI had 6-tiles island

yes I remember in the past that playing again same map was always same start spot

and ye 1-tile island would be even worse start than my archipelago, I had space for 2 cities, in worst case there would be oceans around impassable until compass