AI spawning too close

[hokath]

Perhaps you need to work on your very early game.
Are you settling in place? This isn't always a good idea if it will give you 0 good tiles to work turn 1.
Are you managing your citizens? Sometimes you get pure tiles and pure tiles. It's not good to work only food all the time. Here I grow to pop 2 and then I swap over to work tiles to get the monument out several turns quicker before I go back to growing -- all depends on the terrain you get.

I didn't even see his scout coming it walked right up to my only city.

You said this as if it is scary? Even if you had an undefended worked, AI will not declare war on you this early.
As to the other timings, Turn 14 is perfectly normal to find AI. Scouts can move fast. Turn 34 if anything is a bit late for the second one.

 

[axatin]

Part of the problem I often find is the AI searching around for a good first settling spot. Often moving really close to another AI...or me! I wish I had the modding chops to create a mod that forced everyone to settle in place on turn 1.

That's not difficult to change. Go to "(1) Community Patch/Core Files/Core Values/CoreDefines.sql" and change the 1 in line 3 to a 0:

Spoiler :

 

[detroitrex]

I had a start (think it was 4.13) where on turn 2 or 3 my scout ran into spain's settler. Earliest worker I ever had

 

[azum4roll]

That's why giving Pathfinders base 3 moves is a bad idea

 

[paul_soryu]

I had pretty normal starts last couple of games (comunitas, low sea level), but I noticed AIs on the other continent are crammed to the point that some of them can only settle 3-4 cities in total, while on my continent there is a huge landmass separated by sea (not ocean) without any civ, and also, in one of the games - a big archipelago away from the crammed continent. This would be mitigated if some civ conquered the other, but they were really bad at early-mid game conquering in my games.

 

[L. Vern]

For those curious, I have a video of quite a few replays on V4.5 Communitu

I think one thing that really stands out about the starts is there's quite a few games where the majority of civs start on one continent, I've skimmed through and found a few of the most egregious examples:

Spoiler Skewed Spawns :

The last one in particular here looks really off, it's very difficult for me to believe it's incorrect to spawn at least one or two civs on the other landmass.

Additionally, I've found that even on non-skewed spawns it is very often the case that initial settlers will spawn at the limit of 9 hexes distance, as a random example:

Spoiler Balanced Spawn :

I don't have any hard numbers to back this up (I might implement an average capital distance to closest capital metric) but just from visual inspection of a couple hundred generated maps and spawns I can confirm Communitu standard games are frequently mega-cramped.

Looks to me like a lot of the maps are generated with one of the continents being extremely small and/or waterlogged and is probably a big contributor to the observed close proximity spawns

 

[azum4roll]

Map generation is entirely random (Perlin noise) outside of avoiding land near the map wrap.

Rifts aren't, but IIRC they're drawn starting at a random X coordinate near the middle at the bottom of the map and move upward in a way of trying to erase the least amount of land, without caring about the size of resultant continents.

In other words, the pre-rift maps are natural but rifts are artificial. Would be nice if anyone can suggest a way to make rifts start at a coordinate that will chop the amount of land roughly into half.

Also, we could use a new landmass class that is only separated by ocean and ice, in the DLL.

 

[L. Vern]

Rifts aren't, but IIRC they're drawn starting at a random X coordinate near the middle at the bottom of the map and move upward in a way of trying to erase the least amount of land, without caring about the size of resultant continents.

That's really interesting, had no clue that's how it was generated!

Would be nice if anyone can suggest a way to make rifts start at a coordinate that will chop the amount of land roughly into half.

You could do something like the following for calculating center of mass on the X axis:

function get_rift_longitude:
    sum = 0
    n = 0

    for plot in all_map_plots:
        if plot is land:
            sum += plot.X
            n += 1

    return sum / n

I think this does what you want, though correct me if I'm wrong and maybe we can think of something else

 

[azum4roll]

It'll still need to avoid erasing land. The center of mass could be all high altitude plots and would need to be avoided at all costs.

The map is way prettier without rifts, but the game is unfortunately balanced on having 2+ continents.

 

[KungCheops]

I think it's kind a weird that when one continent is considered "too small" for two civs, the other continent will have all 8 even if it can only have like 5 or 6.

Maybe just squeeze two civs into the other continent instead

 

[azum4roll]

I think it's kind a weird that when one continent is considered "too small" for two civs, the other continent will have all 8 even if it can only have like 5 or 6.

Maybe just squeeze two civs into the other continent instead

That's worse for the two civs.

Terra-style placement at least still makes sure everyone has similar amounts of fertility (not perfect because of islands accessible by coast).

 

[KungCheops]

It's either making the game unplayable for two civs, or making the game unplayable for all civs.

Doesn't matter to me though, I always play with completely unlimited starting placements.

 

[Trumtram]

I think one thing that really stands out about the starts is there's quite a few games where the majority of civs start on one continent, I've skimmed through and found a few of the most egregious examples:

Thanks for the examples. I can totally confirm these. Had multiple games where a whole continent consisted only of city states with Civs being cramped in on different land masses. Good to have these, and hopefully this problem can get solved in the future.

 

[KungCheops]

Thanks for the examples. I can totally confirm these. Had multiple games where a whole continent consisted only of city states with Civs being cramped in on different land masses. Good to have these, and hopefully this problem can get solved in the future.

There is a way to turn off the "must have enough resources/space/whatever" check in the map scripts, don't remember off the top of my head what it's called right now. Tectonics has an explicit option for it.

Although Azum will argue that this makes the game "unbalanced" (it's a single player game) (all map scripts are unbalanced)

 

[hokath]

Looks to me like a lot of the maps are generated with one of the continents being extremely small and/or waterlogged and is probably a big contributor to the observed close proximity spawns

Yes some of these are completely flooded, Atlantis-style continents, very strange I don't ever see these. Guess it's just RNG.
However the last 2 it does look like you could put 2 civs there. Perhaps the fertility threshold could be lowered slightly?

 

[azum4roll]

There's no threshold. Each civ placed decreases the fertility of the landmass/area (depending on placement method) by a pre-determined amount that depends on total fertility of the map and number of players.

Landmasses that have fertility lower than 25% the global amount or 2x the above pre-determined amount will not be considered. Maybe that can be lowered?

 

[hokath]

Yes that 25% is what I'm getting at.
Because if I have 8 civs and 2 of them are on a continent then 25% is the ideal number, it could go to like... 22%? It would be good if we could test. Is this printed anywhere?

 

[azum4roll]

Yes that 25% is what I'm getting at.
Because if I have 8 civs and 2 of them are on a continent then 25% is the ideal number, it could go to like... 22%? It would be good if we could test. Is this printed anywhere?

It's somewhere in ASP.lua. Currently line 1829.

Having this number too low may skew the placement into archipelago mode (arguably worse) more commonly. Maybe make it also dependent on number of players?

 

[azum4roll]

Wait no, it's 25% the fertility of the "biggest landmass" that's the threshold. Should already be fine for 8 players.

Please upload Lua.log if anyone sees Terra being used for Continents-styled maps. It should contain the fertility of each area/landmass.