analog vs. digital signals

Picture a dimmer knob versus a light switch. The dimmer can sit at any brightness — a smooth, continuous range. The switch has only a few fixed positions: on or off. An analog signal is like the dimmer: a value that varies smoothly over time and can take infinitely many in-between levels. A digital signal is like the switch: it holds one of a small, fixed set of values, most often just two — a high level for 1 and a low level for 0.

Concretely, a signal is just some physical quantity (a voltage, a light intensity, a radio wave) changing over time. An analog signal traces a continuous curve, so a sound wave or the original AM radio voltage can take any value at any moment. A digital signal is engineered to stay near a few clean levels and to switch between them at regular instants; a receiver only has to ask 'is this closer to high or to low?' rather than measure an exact value. That makes digital robust: small distortions can be cleaned up by snapping the wobbly received signal back to the nearest legal level, so the original bits survive even after the signal has been knocked about.

This is why nearly all modern networking is digital end to end — errors do not silently accumulate the way they do when you copy an analog tape of an analog tape. But the medium itself is fundamentally analog: a wire only carries voltage, the air only carries waves. So a digital signal is really an analog signal we have agreed to read as discrete levels, and at high speeds it always looks rounded and noisy up close. Modulation (later) even rides digital data on top of an analog carrier wave on purpose.