Snake

On a Nokia 3310, Snake was flawless. Four keys, 2, 4, 6 and 8, each with real travel and a click at the bottom. Your thumb knew where it was without looking, and a turn happened the instant you pressed.

Then phones lost their keys and Snake didn't adapt. It kept its square grid and its four directions, and every modern version I tried simulated the missing keys. In a game where one late turn ends the run, each simulation adds a tax.

1. You eat to grow. Every success makes you longer.
2. Your body becomes the maze. The longer you are, the less room you have.
3. The walls close the space. There's a limit to how far you can run from yourself.
4. Momentum is risk. You're always moving, and going faster makes everything harder.

None of the four rules needs a grid. The grid was how 1997 hardware expressed them, not one of them.

The model

Don't ask which device. Ask which way.

A thumb on glass, a mouse pointer, two keys and an analog stick all describe the same thing: which way do you want to go. The first thing I built, before the snake looked like anything, was the layer that turns each of them into one angle.

The game listens for whatever you touch first and switches as you do: pick up a gamepad mid-run and it follows the stick. The turn limit is what keeps it Snake. If the worm could turn instantly you could always spin away from danger, and your body would stop being a maze.

The camera

High enough to plan, close enough to feel.

The camera floats high and behind the worm on a damped spring, so it lags a beat behind sharp turns and then catches up. It aims a little ahead of the nose, and widens slightly when you boost. The way the view leans and settles as you carve a turn is the moment the game stops feeling like a port. It's the one part I never changed.

The items

Read it before you reach it.

At speed there's no time to look anything up, so each item explains itself by shape: round is safe, a star is a prize, a tail means speed, spikes mean harm. The two in the middle give something and take something back, so going for them is a decision.

The world

My first pass looked cheap, so I wrote the direction down first.

The first space theme was a neon ring on a black disc with heavy glow everywhere. Nothing told your eye where to look. The second time I started from one sentence: a flat-topped shard of rock adrift in deep space, lit by a low sun.

What I learned

Design for the input, not the device.

Four devices became one question, “which way?”, and device detection became a detail. It's how I'd approach any product that has to work on whatever someone is holding. Writing down what the game was before touching it is what made that possible.

What's next

It hasn't been through structured playtesting yet. The next step is watching five people pick it up cold, on their own devices.

1. Terrain. Would hills, ramps or tunnels deepen the steering, or ruin its purity?
2. Other people. One heading per player is tiny to send over a network. Is the game better with friends in the same arena?
3. The other side of the screen. In every Snake the game places the food and you react. What if you placed it, and the worm had to work out how to reach it?

That last one is a small experiment called Architect, tucked away at the foot of the start screen. I don't know yet whether it's a game, a toy or a dead end.