Scaling a game world both for realism and for mechanics

I am currently working on a turn-based 4X game, and one of the things I quickly ran into was a question of scaling. On one hand, I want things like realistic sizes (having some idea of what each tile represents in terms of area), but on the other hand, doing so would limit me in terms of certain mechanics I want to implement (the size of a realistic city would not even occupy a tile, let alone multiple tiles, a mechanic I want to use).

How would I go about scaling the world so that I can have both realistic scaling AND the mechanics I want?

Some things I have considered:

Changing the size of the world. Nothing says that I have to have this take place on an Earth clone (especially with randomly generated terrain). Smaller worlds might allow the detailing I'd need for the cities without having unrealistic tile sizes.

Foregoing the realism: Something that came up while I was researching was that ancient cities were tiny by today's standards. Ancient Rome (the part encompassed by the Aurelian Walls) was only 5.3 square miles (for comparison, Manhattan is 22.96 square miles- enough to fit four Romes). I could forgo the realistic sizes of cities and make them the size I want for my mechanics.

Not including the entire world: Depending on the time-period I end up basing this off of, it may be pointless to include areas that effectively did not exist.

What is the best way to balance realism and mechanics for the scaling of a game world?

• Well what is a tile? If you just have one planet, say the size of Earth, and a 1km by 1km "tile" (not that squares really work if you want an entire planet), that is around 40,000 around the equator or about 500 million total, which depending on your design may be perfectly possible. 2km may be viable as well.
– Will
Aug 4, 2013 at 21:13
• There is no balance, game mechanics must always win. Aug 4, 2013 at 21:18
• @WillNewbery Well, I was thinking of using hexagonal tiles, and 500 million is not feasible unless I can reduce each tile down to a few bytes each. 2 km may work a bit better though. Aug 4, 2013 at 21:23
• Well if the number of tiles that are "special" is very small, then the main tile array could just use say an 8bit tile type. You might even be able to compress it for large regions of desert, sea, etc. Then to deal with special stuff you could have a hash map for coordinates->SpecialObject or something like that. I don't really see a player wanting to pay attention to such details anyway, only caring for cities and such represented by groups of tiles.
– Will
Aug 4, 2013 at 21:36
• Well, one of the issues of having a lot of tiles is that it gets tedious to manage, particularly if you need to, say, move troops across it. Aug 4, 2013 at 21:38

1 Answer

Gameplay is always more important than realism. Realism is a tool to make the game more accessible and easier to understand. When game features work like the player expects them to work in real-life (note: the players expectations are not necessarily realistic), it makes it easier for the player to learn the game. But that's it.

As soon as realism starts to make the game less interesting to play, it has to go.

A good example for this is the Civilization series. It has plenty of acceptable breaks of reality. In the beginning it takes decades for a unit to travel to a neighboring city, to name just one. Is this a flaw? Looking at the huge success of the series, it isn't.

So forget about what would be a realistic scale for your game world, think about what would be a fun scale.

• You want players to have some room to expand, but have them meet and get into conflict with other players quite soon.
• You don't want players to manage too many units, cities and other entities at the same time, because this will be tedious.
• You want players to make meaningful decisions, and not force them to do too much micro-management, especially when it has little or no importance.

To reach these goals, you have to compress the world (move meainingful locations closer together).