← Writing

The Ghost Before the Building

•

I remember Age of Empires 2 less as a game than as a world. A long road that starts in the Dark Age with a town center and a few villagers and ends in the Imperial Age with a walled city. Campaigns that walk through history, every civilisation with its own architecture and units, an empty valley turning into a crowded town over a few hours.

I always played the campaigns alone, and I played the game like a medieval simulation. War came last. First the area around the town center filled up, then the edge of the forest, then the farms spread outward in rings. When I saw an empty patch on the map, I’d think about what belonged there.

Out of that whole game, there’s one small detail I can still see clearly: the preview that appears when you place a building. You pick a building from the menu, a grid appears on the ground, and a translucent copy of the building sticks to your cursor. Over the forest, no. Against another building, no. When it lands on a gap that works it says nothing; it simply accepts. You click, and the ghost becomes a foundation.

Just a preview. But the way I played, the decisions I thought hardest about were always made on top of that ghost.

Distance is time

In Age of Empires 2, a villager who has chopped wood carries it to the nearest lumber camp. If the camp is three tiles from the forest, the villager walks those three tiles there and back every single trip. Hundreds of times over a match. If the mill is far from the farms, the farmers spend their day on the road. A house isn’t just room for five more people; it’s a wall that blocks a path, an edge the next building leans against.

The ghost tells you none of this. All it says is “fits”. But that click decides how many minutes the building will save or cost you for the rest of the match. Once the foundation is down, that is the geometry of your town.

Before the armies, the age-ups and the sieges, this was the real game inside Age of Empires for me: space.

Borders

Then came Rise of Nations, and I played a lot of it. Rise of Nations added one more thing to that question: borders.

The land that was yours was marked by a line. You could only build most things inside it, and armies that pushed into enemy territory wore down the longer they stayed. Every new city or fort you founded pushed the line a little further.

It was a small rule, but it changed what placement meant. In Age of Empires, space was plentiful and distance was expensive. In Rise of Nations, space itself was expensive. Where you founded a city drew the border the next ten buildings had to fit inside.

Put the two games side by side and I see the same thing: my favourite moment in a strategy game is the moment I decide where something goes. War is the consequence of those decisions.

Taking games apart

Since I was a kid, I’ve taken games apart while playing them. Why is this building so expensive? Why is this menu here? What happens if I remove this rule? Making games has never been a someday plan for me; it’s a curiosity that is always running in the background. When I finish a game, what stays with me is usually not its story but one of its rules.

The game I’m working on now came out of that curiosity circling the same question for a long time. It doesn’t have a name yet, but its question is clear: if I strip a strategy game down until only placement is left, what do I have? A game, or a piece of one?

To take that question seriously, I first had to look at how much room placement actually takes up in an RTS.

The share of a decision

In an RTS match, placement takes up very little room. In a forty-minute match you give hundreds of commands: send villagers to resources, train soldiers, scout the map, answer an attack. Placement gets lost among them. Most of the time it happens in a hurry, on the way to the next thing.

RTS match · 40 min One level · 4 min
command placement
Two timelines at the same width. Above, placement decisions get lost among hundreds of commands; below, the level is made of nothing else. The ticks are illustrative, not measured.

What I want is the opposite: shrink forty minutes to a few, and leave nothing in those few minutes but placement. Fewer decisions, but heavy ones. Each slow enough to think about, each important enough to have a consequence.

Doing that risks losing two things. The first is pressure. In an RTS the clock runs and the opponent doesn’t wait; take that away and decisions get light. The second is scarcity. If the map is endless you put everything in its best spot, and then there’s no decision left at all. Something had to take the place of the clock and the opponent.

One sentence

This is the idea in one sentence:

The tension of Tetris’s “where do I fit this?”, with the pleasure of building a village.

Tetris found the replacement for the clock long ago: you don’t choose the next piece. Having to place something you didn’t choose creates tension even in the calmest moment. Scarcity comes from the board itself. The number of squares is fixed, and it shrinks as you fill it.

In this game, buildings are like Tetris pieces. A house is one square, a farm is 2×2, a smelter is a two-square bar, a market is a T, a dump is an L. The board is a valley of squares, as small as 5×5 or as large as 10×12, and its shape changes every level. The shape of the board is half of the level’s puzzle.

Next three
House
Well
Farm
Still to come
House
Smelter
Mill
House
Dump
Market
House
Hold
10 buildings
Attempt 1
Attempt 2 same order
Attempt 3 same order
A level's deck. The order is fixed: every attempt brings the same buildings in the same order. The next three are on screen; one building can be put on hold. The red ones are dirty buildings.

Where I part ways with Tetris is the order. In Tetris the pieces are random. Here each level’s deck is designed up front and stays fixed: every attempt brings the same buildings in the same order. The next three buildings and the number left are on screen.

That decision has a cost and a payoff. A random deck makes every game different, but when you lose you blame luck. A fixed deck turns every level into a puzzle. When you lose you can’t say “I was unlucky”; you say “I put it in the wrong place”. There is a solution, and you can find it. That’s also what the stars mean: one star is passing the level, three stars is solving the puzzle.

I’m not giving up on randomness entirely. Later I’m thinking of a daily mode where everyone plays the same random deck on the same day. But the main path will run through hand-designed levels.

The bad neighbour

The deck isn’t made of good buildings only. The dump comes too, and the smelter.

In city builders you push the ugly building into a corner of the map, or you never build it. That option doesn’t exist here. The deck hands it to you, and it gets a place on the board. Often the real question of a level isn’t where the nice buildings go, but whose neighbour the bad one ends up being.

This rule sets the tone of the game, too. You’re not building a perfect village; you’re building the best village you can with what you were given. The dump is going to be there. The question is how far the houses stay from it.

Score comes from neighbours

A building’s value depends on what’s next to it. “Next to” means the eight surrounding squares; touching at a corner counts.

Leaving out diagonals would be a cleaner rule on paper, but it’s hard to read by eye. When two buildings touch at a corner, the player sees them as neighbours. A rule shouldn’t contradict what the eye sees.

In the first draft a house scores like this: +2 for every neighbouring well, −3 for every neighbouring dirty building.

F W D H F W D H

+2 Total +2

+2−3 Total −1

The same house, one square apart. On the left it touches only the well; on the right, the well and the dump. W well · D dump · F farm · H house.

There are two subtle rules inside that sum.

First: a neighbouring building counts once, however many of its squares touch you. A house touching two squares of a farm sees one farm, not two. Otherwise big buildings would carry unfair weight with their neighbours, and the board would favour whatever has the largest shape.

Second: terrain, on the other hand, counts per square. A lumber camp touching three forest squares scores three times. Among buildings the goal is finding the right neighbour; with terrain the goal is burying the building in it. Two different questions, two different counts.

Then there are rows. In Tetris, a full row clears. Here every full row and column is worth +5, but it doesn’t clear.

Tetris

row clears

This game

row stays · +5

In Tetris a full row clears and the board empties. Here every full row and column is worth +5 but stays where it is; when the level ends, a village is left on the board.

This may be where the game parts most from Tetris. In Tetris, success is emptying the board. Here success is filling it, and what you fill stays. When the level ends, you’re left not with a score but with a village.

What the ghost says

The ghost in Age of Empires only told you “fits” or “doesn’t fit”. In this game it has to say one more thing: “Fits, but it’ll cost you this.”

While you drag, a +/− appears above every building whose score would change. The player doesn’t do the maths; they see the result. Move the house around the board below; the best spot is +4.

W W SM D F House +2 +2 −3 −3 −0
Total
Hover over the board or tap a square to move the house. Arrow keys work too. W well · SM smelter (dirty) · D dump (dirty) · F farm

I do this to stop memorising rules from being a reward. A player who has memorised the rules shouldn’t beat one who hasn’t; the edge should come from finding the good spot. Hiding information isn’t difficulty, it’s friction. The difficulty belongs on the board, not in a menu.

I took one more rule from Tetris: hold. You can set the incoming building aside and take the next one, but only once per new building.

There’s no undo, though. If you could undo, no placement would be a real decision; you’d drop the ghost here and there, keep the best and take back the rest. Hold gives you options without erasing the decision.

A level is lost in two ways: when no move fits, not even with the hold, or when the deck runs out with the score below the target. A level lasts three to six minutes.

Why the phone

Strategy games usually ask for a desk, a mouse and an evening. The phone is more accessible. But accessibility isn’t only about the device; it’s also about time.

Because there’s no clock, the game can wait for you. When the bus pulls in you put the phone in your pocket and pick up where you left off at the next gap. Nothing burns down, nothing is missed. You can’t walk away from an RTS halfway through; you can walk away from this.

The interaction comes down to one motion, too: drag, rotate, drop. With one finger. The ghost in Age of Empires was dragged with a mouse; on the phone it’s already under your finger.

That comes with a constraint: the board has to fit on one screen, and every square has to be big enough to pick with a finger. It puts a ceiling on the size of the board. Good. A small board is one where every square matters.

The rules I started with

Before drawing a single building, I wrote myself a few rules. Every new idea goes through them first.

No copies. With every new mechanic, the first question is: does another game already do this well? The game’s close relatives are clear: the placement tension of Tetris and the neighbour scoring of Islanders. In Islanders you choose buildings from one of two packs; in Tetris the pieces are random and time runs. This game differs in three ways: you can’t choose the piece, the bad buildings are mandatory, and every level has a solution. A mechanic that adds nothing to those three doesn’t get in.

No pressure. No lives, no energy, no timers, no ads you’re forced to watch. When you lose a level, you retry right away, for free. In mobile games these patterns exist to keep players away from the game and sell the distance back to them. In a puzzle game, retrying is the learning. Making it expensive makes learning expensive.

Mechanics first, theme second. The theme is put on top of a mechanic afterwards. If a group of levels teaches a new obstacle, those levels take place wherever that obstacle feels most natural. If the obstacle is something that melts, the levels happen in the snow.

Few words. No story text. A short card to introduce a new building, a single line when you enter a new place. Long text takes the eye away from the board it should be on.

Say it without shouting. No numbers flying across the screen, no “NICE!” or “AMAZING!”, no confetti on every move. When you find a good spot you already know it: the neighbours’ scores change, a row fills up. The game should say it, not shout it.

For now

What I have is a page of rules and a table of buildings. The numbers are all a first draft; they’ll change as the game gets played. What won’t change is the question: in a game where you don’t choose what comes, how do you decide where things go?

In the next part, I’ll write about what this valley looks like.