[GS] Thinking about optimal city spacing with in-game sample

[Socrates99]

I wouldn't mind much Civ specific differences as Civs are supposed to de different and differently strong.

I thought, that having on average per city more tiles at your disposal may lead to another "optimal" mix of tile usage type in general. And that somewhere in the occult AI code might be some Magic Numbers, which refer to a sought distribution of tile usage depending on the standard range 3. And that perhaps the weight of some correlated Magic Numbers should ideally also be adjusted hand in hand ...

Otoh, as we have the best publisher in gaming, I'm sure we can have a look onto the AI code soon.

I mean, you could see exactly what their cities are likely to work by just leaving your own cities auto assigned. It's not great. Having extra tiles is something everybody gets and players win when it comes to stuff everybody gets.

 

[FlyinPhil]

This makes me question, when considering overall gameplay, is it actually better to aim for an end state of a city

I am ALWAYS thinking about end state, for two reasons. The first is useless in the context of gameplay, as I am a bit of an aesthetic player, and I like to build 'beautiful' cities with a lot of features and optimizing certain things and even leaving a lot of unmodified hexes in between cities.

The second is that, yes, all the time hexes go unused between cities over the course of the game represents untapped yields (be it production, food, gold, etc.), but the utility of YOUR hexes towards the end of the game is also a function of the victory condition you are seeking. If I'm playing a domination or religious victory, then more important to me are my rivals' hexes, and how many of them I have 'occupied' by the end of the game. It doesn't matter if my own cities are small and have limited utility, as long as I'm pillaging or eliminating entire religions when I sweep across enemy territory I'm getting what I need to win. However, if I'm playing a science or cultural victory (dependent on yields in one case, and appeal or space for wonders on the other), then I'll need those cities to have expanded towards the end game. In particular, when playing a science victory, I will scope out 3-4 sites for large cities, usually including my capital, that I can max out production in for building the laser stations, and leave more open space for those relative to my others.

When I first started playing Civ 6, I always tried to max out city spread, so no less than 6 hexes between cities (every city got all three rings). I realized this was VERY suboptimal from a gameplay perspective, so I adjusted down to a default of 5 hexes in between (sharing the third ring), and now that I've been playing for over four years since the game came out, I'm actually down to a default spacing of 4 hexes in between (so basically every city gets two rings worth of hexes).

Another technical consideration? As Victoria said:

every move you make with a settler before settling is in itself less optimal before you consider the benefits of the target settle

So my question is, do you play on a small map on fast speed? Or do you play on a large or huge map on marathon speed? I play marathon, which means each turn I spend with a settler in transit costs me less time than it would on a standard or fast game.

The bottom line: it's all dependent. I will flex the spacing for each and every city based on my needs and where I want the city to be when I seek my victory condition.

 

[uhu]

Thinking about optimal [] spacing is a great starting point for the experience of transcendence.

I mean, you could see exactly what their cities are likely to work by just leaving your own cities auto assigned.

'another "optimal" mix of tile usage type' meant the distribution between improvements, districts, wonders and letting unused. This value changes with the number of tiles available on average per city. I'm quite sure, that the game doesn't use this value Explicitly - but Implicitly! - and changing the rules is one thing (game engine and AI players will abide to the rules ofc), but indirectly changed rules weaken the AI players, because of slightly (more) mis-adjusted weights.

I'm actually down to a default spacing of 4 hexes in between (so basically every city gets two rings worth of hexes).

How is "spacing of 4 hexes" resolved in the second dimension? Seems very tight.

Eg this picture shows in the upper part a regular grid with distance 6 5 while cities share their 3rd rings completely. All cities receive on average 6 + 12 + (6/3 + 12/2) = 26 tiles.

The lower part shows, how more tiles in "partly doubled ring 3" can be gained by using ribbons of tiles in one dominant direction. Think of it extended the same way to the left and right as well as on the upper side.


edit: corrected distance 6 to distance 5

 

[FlyinPhil]

How is "spacing of 4 hexes" resolved in the second dimension? Seems very tight.

I think we have a difference of definition. When I say four hex spacing, I think I'm referring to what you are calling a distance of 5. What I mean by 4-hex spacing is 4 blank hexes followed by a city center on the 5th hex (see my diagram below). Your grid is what I would call 5-hex spacing, and I still use that typically for my capital and other key cities.

All cities receive on average 6 + 12 + (6/3 + 12/2) = 26 tiles.

This used to be my standard placement, but even on long science victory games, I find there just isn't enough time for most of my cities to use the entire space, and my civ's overall productivity is increased by squeezing in more city centers than it is by expanding the borders of each individual city. I would say my average city size by end game is about 12 (usually 20+ for some core cities, 15ish for the next group, then 6-8 for late-stage cities). You could also optimize this by settling initial cities further apart, then settling later cities closer together.

Here's another way to visualize 4-hex spacing. By using straight spacing (on the right) rather than jagged spacing (on the left), you get 6 more hexes per city. My patterns are usually a mix of the two, based on the space on the map and where key hexes are (wonders, resources, etc.).

EDIT: every once in a while, I might do straight 6-hex spacing for a capital. This means 36 hexes within the capital's usable range, but that usually means there's some water or mountains that are taking up some of that space, or I plan to build a lot of wonders, or I really think I can get the capital to ~30 population (and thus actually use most or all of those hexes).

 

[Victoria]

each turn I spend with a settler in transit costs me less time than it would on a standard or fast game.

Sure, I play standard everything and normally continents and islands to allow for whatever civ I get. There is still loyalty but don’t get me wrong, it’s a question of what suits the civ and situation. Playing Kupe or Victoria I settle cities all over the place..... but victoria deity I’ll settle the first 3 within 4 tiles purely for loyalty, she often goes dark early.
I often settle a city within 4 tiles of another city purely to use some of the other cities tiles for a boost, especially food tiles.

 

[uhu]

I think we have a difference of definition.

More simple: I just goofed up. The unmodified game uses for city spacing a GlobalParameter 'CITY_MIN_RANGE' with the default value of 3. I think, we should stick to this definition of distance / spacing. (As you did.)

Here's another way to visualize 4-hex spacing.
View attachment 596574

If we add in the compressed variant a pink "berry" right of the blue, I suppose, this 4 berry pattern can be repeated endlessly in all directions?!?
Hhmm, would that lead to 17 1/4 tiles / city on average, since pink and beige berries are overlapping by 3 tiles per pattern?

I'm still curious, how a practical continuous "spacing of 4 hexes" grid looks like.

I often settle a city within 4 tiles of another city purely to use some of the other cities tiles for a boost, especially food tiles.

Yeah, switching extreme food tiles between 2 cities to grow one rapidly while the other "currently stagnating" is heavy on production. The near settlements are often also 1 turn earlier to reach for the initial settler and subsequent builders and defenders. All in all very efficient - rapid return of investment

Btw, you can build the most compact valid grid with a "spacing of 3 hexes" in the above picture by leaving out the green dots and overlapping the blue dots by building a similar web of shared outer rings. But all cities receive on average only 6 + (6/3 + 6/2) = 11 tiles.

 

[FlyinPhil]

If we add in the compressed variant a pink "berry" right of the blue, I suppose, this 4 berry pattern can be repeated endlessly in all directions?!?

You can actually fit them together with no overlap (picture where the beige "berry" is relative to the green "berry", and you can place your pink "berry" the same relative to the blue "berry"). It could still repeat ad infinitum with 18 hexes per city.

 

[uhu]

Yep, looks like you can have 3 tiles less or 2 tiles more per pattern.

 

[FlyinPhil]

Yep, looks like you can have 3 tiles less or 2 tiles more per pattern.
View attachment 596666

If you drop that second cluster down one (such that the bottom row continues slightly diagonally), you could fit a pink "berry".

I'd be impressed by a map that could fit this pattern for more than a few cities! Ultimately, it's a little silly to imagine a scenario where you could have this perfect of a pattern, but for inexperienced players like the OP, knowing the geography of Civ cities is a key skill.

 

[aieeegrunt]

Civ VI is quantity over quality so the honest answer is to pack em in like sardines. The only caveat might be CV for land based tourism (wonders, UI, parks) but even a crowded empire could use great works and rock bands to a similar effect.

I used to play like that. Pack my cities tight. Build just a few districts per city. Maximize trade route capacity, spam my victory district, build a few prime campuses, IZs and EDs. If a city's production is weak use Reyna or Moksha to buy the district and just buy the buildings.

Too much min maxing, too much planning, kind of boring and way too much tile counting. So I went with a mod that increased the min distance by 1 and really enjoy it. I still pack them in, a habit I can't seem to kick. Theres usually good tiles to work and plenty of room for wonders and districts. I like it.

Quantity over quality is EXACTLY how to descrive Civ6 sadly. You are always better off having more cities because that means more districts, and the simplistic board game mechanics favours that over better cities

I am at loss to understand why this game’s design seems to think it only has cardboard counters and paper maps to work with