What each controller felt like

The joystick: four switches and a plate

A typical arcade joystick measures nothing. Around the bottom of its shaft sit four switches, for up, down, left and right. Push the stick and it clicks one switch closed; push it toward a corner and it closes two, which the game reads as a diagonal. So the stick knows nine positions: eight directions and the middle. The gate, a plate around the shaft, decides how the stick moves between them. An eight-way gate guides you into the corners. A four-way gate blocks them entirely, which suited maze games, where a corner pressed by accident could steer you into a wall.

The trackball: a ball with weight

A trackball is a heavy ball resting on rollers, one for each direction. As the ball turns, each roller spins a slotted wheel between a tiny light and a sensor, and the game counts the flickers. Because the ball has weight, a hard flick keeps it rolling after your hand leaves it. Good players learned to spin it fast across the screen and stop it dead with a palm. Centipede (1981) and Marble Madness (1984) were built around that feeling.

The spinner: a knob with no end

A spinner is a knob that turns forever in either direction. It reads movement with a slotted wheel, like a trackball, and some had a small flywheel inside to give them weight. It is different from the older paddle, a dial that stopped after most of a turn and reported its exact angle. A spinner only reports how far it has moved, so it suits anything that slides along a line or runs round a ring: the bat under a wall of blocks in Arkanoid (1986), or a shape crawling round the rim of a tunnel in Tempest (1981).

The aim controller: the screen gives itself away

Arcades and toy shops called this kind of controller a light gun; we call it an aim controller here, and the computer version was a light pen. It cannot see a picture the way you do. It has one small light sensor behind a lens that narrows its view to a spot. The trick is in the screen. A picture tube draws its image with a beam that sweeps every row many times a second, so each spot on the glass lights up at an exact moment. Some controllers timed that moment to work out where they pointed. Others made the game blank the screen for one frame and draw each target as a white box, then checked whether the sensor saw light. We show you that frame as a still picture instead of flashing it.

Both methods need a tube. A flat panel has no sweeping beam, and it shows each picture a moment late, so the old controllers get nothing useful from it. That is why the originals do not work on a modern television.

The control comes first

A maze game on a trackball would feel slippery, and a tunnel game on a joystick would feel stiff. Designers often chose the control first and invented the game that suited it.