Anyone know how it chooses what tile to expand to?

[Victoria]

Because there is no sound reason why it should choose these tiles

 

[iammaxhailme]

It definitely prioritizes tiles with resources on it over tiles that don't, regardless of yield. Besides that, no idea. It picks poorly.

 

[Victoria]

Well at least you explained why it's doing it, thanks.

 

[WillowBrook]

And I don't know if we can do things that have an effect. I've noticed several times the pink hex jump around during the countdown to the next expansion. That could either be due to a player action affecting the valuation of the tiles, or because it randomly picks when multiple tiles have the same priority and something reset the random choice.

Anyone know if there's anything we can examine under the hood? I don't know of modders have found anything.

 

[iammaxhailme]

I think I was wrong earlier...

https://steamuserimages-a.akamaihd....014/77F8A37FD5D82F76F6AF176783EAB38953CB0A27/

 

[antimony]

I think the resource prioritization definitely occurs within the same ring, but in that case it needs to complete the 2nd ring before getting that incense. (Note: I am not 100% sure it needs to complete the 2nd ring before grabbing tiles from the 3rd.)

One thing I found frustrating (for city planning) in the past is that the game would sometime change its mind on the next tile for expansion, for no apparent reason. This is anecdotal, but I'm curious if others also got that behavior.

 

[Victoria]

This is anecdotal, but I'm curious if others also got that behavior.

I have definately seen this, but not since the patch yet

 

[kaspergm]

I have seen this after the patch. It did it just the other day.

I have no hard evidence how the tiles are chosen, but everything I've seen indicates to me that it works under the same basic algorithm that was used in Civ5, just worse tuned (or I might be remembering Civ5 in a better light because I always used a mod that let me choose which tile to expand to, which would be an essential mod for civ6 also). Anyway, in Civ5, it valuated the tiles based on their yield and distance to city. I don't remember if it took resources directly into calculation or only indirectly through yields, but above example seems to indicate that Civ6 values resources directly. I've also seen that Civ6 seems to put (too) much weight on distance, so that it will grab worthless ocean tiles in second ring before it takes valuable tiles in third ring which is a real nuisance.

 

[Vahnstad]

1. tiles with luxury resources
2. tiles with resources
3. land tiles above ocean tiles
4. cheaper tiles (distance)

 

[Vahnstad]

I think I was wrong earlier...

https://steamuserimages-a.akamaihd....014/77F8A37FD5D82F76F6AF176783EAB38953CB0A27/

The tiles with resources cost 50 coins, so it will first take all the tiles with 30 coins (ocean), so if you want to improve the 50 tiles sooner than later, then you will have to buy them. In this case, i would buy the 2 incense tiles, and the tile with that hill next to Kilimanjaro, the wheat and the incense. The hill in the southeast is also a good option to buy. But with Indonesia, the coastal tiles are of course also valuable. Early worker is vital here when you have Ilkum (maybe two workers), try to have that build 5 farms eureka at least, and then switch to serfdom and build more builders.

Good city actually.

 

[Rosty K]

In my experience it doesn't always end up expanding to the tile it shows in purple... I'd even say most of the times it picks a different tile at expansion...

 

[agonistes]

Since patch, based on my observations, any water resource gets top priority. Then any land resource/natural wonder.

What grabs my goat is when it picks a 2/0 or 1/1 over a 2/2. It often will pick a 2/1 jungle over a 2/2 jungle.

Ultimately, I really don't care I guess. Maybe I even appreciate the added decision-making it creates.

 

[UWHabs]

Definitely agree that it seems worse than 5. While obviously it can't understand that I want it to expand to a random desert tile because that's where I want my campus to go, I would still in general rather have it expand to 3rd tile resources instead of 2nd tile "plain tiles". If I were doing the algorithm, I think I would keep it simple:
1. Of the cheapest tiles, if any luxury/strategic resources, go to them
2. Of the cheapest tiles, if any bonus resources, pick them
3. Of the cheapest tiles, if any "special yields" (ie. natural wonders or adjacent), choose them
4. Repeat the above 3 steps, except for any tiles it can expand to
5. if none, then claim the best tile among the cheapest ones, giving preference to any that will "fill in" borders (always annoys me how there's that one random tile in the middle of my empire that takes forever to get filled in), or to ones that are adjacent to luxury/strategic resources

Simple enough, should lead to claiming the "best" tiles more often than not, and while it certainly may not be exactly what you always want, would certainly lead to less "WTH" moments when looking at the next tile.

 

[nzcamel]

I think the resource prioritization definitely occurs within the same ring, but in that case it needs to complete the 2nd ring before getting that incense. (Note: I am not 100% sure it needs to complete the 2nd ring before grabbing tiles from the 3rd.)

I have no hard evidence how the tiles are chosen, but everything I've seen indicates to me that it works under the same basic algorithm that was used in Civ5, just worse tuned (or I might be remembering Civ5 in a better light because I always used a mod that let me choose which tile to expand to, which would be an essential mod for civ6 also). Anyway, in Civ5, it valuated the tiles based on their yield and distance to city. I don't remember if it took resources directly into calculation or only indirectly through yields, but above example seems to indicate that Civ6 values resources directly. I've also seen that Civ6 seems to put (too) much weight on distance, so that it will grab worthless ocean tiles in second ring before it takes valuable tiles in third ring which is a real nuisance.

From memory V would expand beyond the current ring to get a resource; but in VI whatever is the innermost ring is always first priority. It does that 100% of the time. I don't mind that, as it can be thematic either way; but it is frustrating re the tiles it often prioritises within that closest ring!

 

[Archon_Wing]

I love it when it chooses a mountain next.

 

[Victoria]

based on my observations, any water resource gets top priority

This one does not fit that

So ring 2 tiles first regardless of quality
Looks like bonus (any maybe strategic) is first choice with a preference for water over land
Then probably luxury resources

or does ot not always follow the same rules?
I feel some testing coming on

EDIT: This test done 4 times with the same tile order each time

 

[Art Morte]

You might be able to abuse it if you could choose the tiles. Grow three tiles into one direction with the first two tile expansions to block a route. Maybe if you had to complete the 2nd ring before moving to 3rd ring it could work.

And by the way, the pink indicator is totally random, it may indicate a certain tile one turn before expansion and pick another tile when it expands.

What annoys me far more than tile expansion is that if you don't want to commit to a theater district, the monument is the only building to give you tile expansion points, i.e. culture.

 

[Victoria]

What annoys me far more than tile expansion is that if you don't want to commit to a theater district, the monument is the only building to give you tile expansion points, i.e. culture.

4 luxuries, Pantheons/beliefs, Culture CS, trade routes to culture CS or cultured city, natural wonders and pushing the eurekas more than normal.
I know they are not buildings but will expand the capital or city.

Seems to be random

EDIT: But then I shut everything down, start it up again with a few more resoiurces and it still chooses the Niter

A few more tests later and it seems there is a random element decision with weighting.

 

[Takfloyd]

One thing I know for sure is that it will never pick a third ring hex until all the second ring ones are taken. Even if it has to take mountain tiles on the second ring while strategic resources are available on the third.

 

[nzcamel]

EDIT: This test done 4 times with the same tile order each time

Horses before stone - good.
Why crabs before pearls though?