Decoherence & Noise
A qubit’s quantum-ness quietly leaks into its surroundings over time — so you can only run so many gates before the answer turns to noise.
You finally have a qubit — isolated, cooled, spinning in a perfect superposition. You walk away for a single millisecond. When you come back, the delicate blend has quietly dissolved into a plain, random bit. Nobody touched it, yet the quantum magic leaked away on its own. This silent decay — decoherence — is the single biggest obstacle between today’s lab machines and a useful quantum computer. Hold the question: why does a qubit’s quantum-ness fade by itself, and what exactly does that cost you?
Decoherence: quantum-ness leaking away
No isolation is perfect, so the environment is always weakly “peeking” at the qubit — a stray photon, a faint vibration, a wisp of heat. Each tiny interaction acts like a partial measurement, and bit by bit the superposition’s delicate structure bleeds out into the surroundings until it’s just ordinary randomness. The qubit decoheres — it forgets it was ever quantum. Crucially this happens on its own, over time, even if you never deliberately touch it.
Two ways a qubit goes wrong (plus sloppy gates)
Decoherence shows up in two main flavors. Dephasing: the qubit loses the precise +/− phase relationships that interference depends on — the very thing that made it useful blurs out. Relaxation: the qubit leaks energy and slumps from 1 toward 0. On top of that, every gate is a real, imperfect pulse, so each operation adds a little control error, and every measurement can misfire. None of these errors are dramatic on their own — but they accumulate.
Coherence time caps how much you can compute
A qubit stays usable only for its coherence time, so you can fit only so many gate operations in before noise swamps the signal. Deeper circuits mean more accumulated error, and past a point the output is indistinguishable from garbage. This is exactly why you can’t just run Shor on today’s machines: the circuit is far deeper than the qubits survive. Decoherence and gate error together set a hard ceiling on useful circuit depth — and that ceiling is the reason error correction (next lesson) isn’t optional.
A melting ice sculpture. You carve something gorgeous and intricate — that’s your superposition — but it starts melting the instant it exists, and no room is cold enough to stop it completely. You have a short window to do your delicate work and take the photo (measure) before the fine detail slumps into a puddle (decoherence). Worse, every time you reach in to reshape it (apply a gate) you melt away a little more. The art is finishing the whole carving and photographing it before the detail is gone.
Why a real circuit runs out of room (illustrative numbers): 1. Prepare a qubit in a crisp superposition at time zero. Say its coherence time is about 100 microseconds. 2. Each gate takes ~0.1 microsecond and is ~99.9% accurate (about 0.1% error each). 3. Run 100 gates: even ignoring the melt, error compounds — roughly a 10% chance the result is already corrupted. Run 1,000 gates and an error is nearly certain. 4. Meanwhile decoherence is dissolving the superposition regardless; past ~100 microseconds the blend is mostly gone whether or not you’re using it. 5. So usable circuit depth is roughly coherence-time ÷ gate-time, minus the toll of accumulating gate errors — a few hundred to a few thousand operations on good hardware. Shor on a real key needs vastly more. That gap is why raw qubits aren’t enough.
This is the reading. The interactive version — active-recall quiz, a hands-on experiment you run in your own AI, and an earned mastery check — is free in the app.
Start this lesson free →