The Quantum Investor's Notebook logo The Quantum Investor's Notebook
← Back

Sunday, June 28, 2026

The Random Telegraph Signal: Why a Good Qubit Won’t Stay Good

A single microscopic flaw, flipping between two states, is enough to make a qubitThe basic unit of a quantum computer. Like a 'bit' in a normal computer, but instead of being only 0 or 1 it can be 0, 1, or a blend of both at once.’s quality wander from one hour to the next.

What a telegraph signal is

A random telegraph signalNoise that jumps at random between two fixed levels, like a telegraph key clicking on and off. In a qubit it is the fingerprint of one microscopic flaw switching back and forth. is the simplest kind of noise. Some quantity, a voltage, a charge, a magnetic field, sits at one level for a while, then jumps abruptly to a second level. On and off, like a telegraph key tapped at random. It is not the smooth, fuzzy hiss most people picture. It is discrete: two values, and nothing in between.1

Not a smooth hiss. Two values, and a coin flip deciding when to jump between them.

Behind one telegraph signal is one flaw: a single fluctuatorA single microscopic flaw, such as a trapped electron or a shifting atom, that randomly switches between two states and produces one telegraph signal., some atom or trapped electron with two resting places it hops between. Two numbers describe it, the average time it dwells in each state before flipping. From those two dwell times, together with the jump size, you can predict the noise signature of that one fluctuator.1 One flaw at one rate is almost manageable; you could measure it and design around it.

From one flaw to a fog of noise

The trouble is that a real qubit does not sit near one flaw. It sits in a material full of them. Stack up many fluctuators, fast and slow together, and their telegraph signals blur into 1/f noiseA pervasive low-frequency noise whose power grows the slower you look, built from many telegraph signals flipping at different rates. Also called flicker noise., a hum that grows louder the slower you look.2 Physically these flaws are two-level systemsA microscopic flaw in a qubit's materials with two resting states it hops between, the usual source of telegraph and 1/f noise. Often shortened to TLS., tiny imperfections in the glassy oxides and rough interfaces every solid-state qubit is built from.3

This is why low-frequency noise is so stubborn. A fast, even hiss you can average away. But 1/f noise piles up at the slow end, exactly the timescales over which you want a qubit to hold still, so averaging and simple filtering never quite clear it.2

Two stacked time traces. The top is a square wave jumping at random between two levels, one fluctuator switching states. The bottom is the qubit's lifetime (T1), high while the top is in one state and dropping in red when it flips, showing the qubit's lifetime drop.
Figure 1: One flaw, and the qubit follows it. As a single fluctuator flips between two states (top), the qubit’s lifetime lurches with it (bottom), dropping the moment the flaw switches.

Why a benchmark won’t sit still

Here is what that does to real hardware. In 2018 a group at Google watched a single flaw drag a qubitThe basic unit of a quantum computer. Like a 'bit' in a normal computer, but instead of being only 0 or 1 it can be 0, 1, or a blend of both at once.’s lifetime up and down over time. Its energy-relaxation time, how long the qubit holds its excited state, would not hold a steady value; it jumped as a nearby two-level systemA microscopic flaw in a qubit's materials with two resting states it hops between, the usual source of telegraph and 1/f noise. Often shortened to TLS. shifted in and out of tune.4 A qubit could look excellent for minutes, then one microscopic click could sharply reduce its lifetime.

Measured this morning, gone by lunch: one click can sharply cut a qubit’s life.

So a fidelity you quote today is a snapshot, not a guarantee. The coherence timeHow long a qubit holds its delicate quantum state before noise scrambles it. Longer coherence means more time to compute. you publish and the one running your circuit an hour later need not be the same number.4

What the drift costs

This is the part that matters to anyone betting on the hardware. A drifting qubit forces a machine to spend its time re-calibrating instead of computing, and it makes a single gate fidelityA score (for example 99.9%) for how accurately a quantum gate does what it is supposed to do. Higher means fewer errors. figure mean less than it looks, because the figure will not sit still. Worse, many benchmark and error correctionTechniques that combine many shaky physical qubits into fewer reliable ones, so a long calculation stays correct. assumptions work best when noise is reasonably steady; telegraph noise that wanders on long timescales makes that harder.2

Engineers fight back by parking the qubit at a sweet spotA setting where a qubit's frequency barely responds to a given noise, so slow drift disturbs it far less., a setting where its frequency barely responds to the flipping, and by chasing cleaner materials with fewer flaws. But you cannot shield against a flaw that lives inside the chip. Until the materials get quieter, the random clicking of microscopic flaws will keep setting the ceiling on how good a qubit can be, and how long it stays that good.

Sources

  1. Machlup, S. “Noise in Semiconductors: Spectrum of a Two-Parameter Random Signal.” Journal of Applied Physics 25, 341–343 (1954). DOI: 10.1063/1.1721637.
  2. Paladino, E., Galperin, Y. M., Falci, G. & Altshuler, B. L. “1/f Noise: Implications for Solid-State Quantum Information.” Reviews of Modern Physics 86, 361–418 (2014). DOI: 10.1103/RevModPhys.86.361. Preprint: arXiv:1304.7925.
  3. Müller, C., Cole, J. H. & Lisenfeld, J. “Towards Understanding Two-Level-Systems in Amorphous Solids: Insights from Quantum Circuits.” Reports on Progress in Physics 82, 124501 (2019). DOI: 10.1088/1361-6633/ab3a7e. Preprint: arXiv:1705.01108.
  4. Klimov, P. V. et al. (Google AI Quantum). “Fluctuations of Energy-Relaxation Times in Superconducting Qubits.” Physical Review Letters 121, 090502 (2018). DOI: 10.1103/PhysRevLett.121.090502. Preprint: arXiv:1809.01043.