Quantum, one clear idea at a time.
Eight small steps from your first qubit to a better understanding of the headlines. No maths background or account needed.
Begin with a qubitA qubit and a measurement
A qubit is a quantum system with two chosen basis states, labelled 0 and 1. Its state predicts measurement probabilities. Measuring in the 0/1 basis returns one classical result, not the complete state description.
A concrete example
Prepare a qubit, apply operations, then measure. Record 0 or 1. Repeating the experiment means preparing the state again.
Keep in mind: A qubit is not a bit with an extra readable value called “both”.
Check your understandingDoes one measurement reveal the whole state?
No. It gives one outcome in the chosen measurement basis.
Amplitudes become probabilities
A pure qubit state has two amplitudes. They can carry signs or complex phases. Each outcome probability is the squared magnitude of its amplitude. The probabilities add to one.
A concrete example
Amplitudes √3/2 and 1/2 give probabilities 75% and 25%. These are predictions for repeated preparations, not guaranteed counts in a small sample.
Keep in mind: An amplitude of −1/2 means a 25% probability, not a negative probability.
Check your understandingIs an amplitude the same thing as a probability?
No. Take its squared magnitude to get the probability for the specified measurement.
Why phase matters
The plus and minus states both give 0 and 1 with equal probability in the 0/1 basis. Their relative phase differs. Apply a Hadamard gate before measuring and that difference becomes visible: plus gives 0, while minus gives 1.
A concrete example
Plus: (|0⟩ + |1⟩)/√2. Minus: (|0⟩ − |1⟩)/√2. The minus sign changes how the two amplitudes combine in a later operation.
Keep in mind: Multiplying every amplitude by the same global phase does not change observable outcomes. Relative phase is the distinction that matters here.
Check your understandingAre two states necessarily equivalent if their 0/1 probabilities match?
No. A later operation can reveal a difference in relative phase.
Sources: limitations on quantum information
Change the phase. Change the answer.
Start in 0. A Hadamard gate, labelled H, creates the plus state. A second H returns the state to 0. Insert Z between them and the final result becomes 1. Z changes relative phase, so the last H combines the amplitudes differently.
A concrete example
Without Z: 0 → H → H → 0. With Z: 0 → H → Z → H → 1. The final results are exact in this ideal model.
Keep in mind: Do not measure between the gates in this example. An intermediate measurement changes the experiment.
One gate changes the result.
Toggle Z between two Hadamard gates. Follow the state, then read the final measurement probabilities.
The second H reverses the first. The final state is 0.
Follow every state and probability
| Step | Amplitude 0 | Amplitude 1 | P(0) | P(1) |
|---|---|---|---|---|
| Start in 0 | 1 | 0 | 100% | 0% |
| After the first H | 1/√2 | 1/√2 | 50% | 50% |
| No middle gate | 1/√2 | 1/√2 | 50% | 50% |
| After the final H | 1 | 0 | 100% | 0% |
Model limits: one ideal qubit, no noise and no hardware connection. Intermediate probabilities describe the state; no measurement occurs between the gates. Measuring there would change the experiment. The static example above remains readable without this control.
Check your understandingWhat changes when you insert Z?
The relative phase changes. The final H turns that difference into a different measurement outcome.
Sources: quantum mechanics basics · limitations on quantum information
Entanglement connects outcomes
Entangled qubits must be described jointly. For the Bell state (|00⟩ + |11⟩)/√2, measuring both in the 0/1 basis gives 00 or 11 with equal probability. Each local result is random. The correlation cannot be used to choose and send a faster-than-light message.
A concrete example
The ideal Bell-state probabilities are 50% for 00 and 50% for 11; 01 and 10 have probability zero. Each person still sees an unpredictable 0 or 1.
Keep in mind: Matching results in one basis alone are not proof of entanglement. Ordinary shared randomness can also produce matching results.
Check your understandingCan one person choose their result to send a bit instantly?
No. The result is random. Comparing the two records requires ordinary communication.
Sources: multiple quantum systems
Read your first circuit
Read a circuit from left to right. Each line follows a qubit through a sequence of operations. Gate symbols show what happens; measurement produces classical data. A circuit is an instruction diagram, not a photograph of wires inside a machine.
A concrete example
Initial state 0 → X gate → measurement in the 0/1 basis → result 1. X exchanges the 0 and 1 basis states in the ideal model.
Keep in mind: A line in a circuit does not mean the hardware has a literal wire carrying a tiny qubit through a row of boxes.
Check your understandingWhich step produces the classical result?
Measurement. State preparation and gates set up the probabilities for that result.
Sources: circuits · quantum mechanics basics
Random outcomes are not always errors
A perfect quantum experiment can still have random outcomes. Hardware errors are different: unwanted interactions, imperfect operations and inaccurate readout make results depart from the intended model. Error correction encodes logical information across physical qubits and uses checks to help protect it.
A concrete example
A 50/50 ideal prediction does not promise exactly 50 zeros in 100 trials. Finite sampling varies even before hardware imperfections enter the picture.
Keep in mind: A logical qubit is not automatically error-free. Protection has limits and resource costs. Reliable computation also needs reliable operations on encoded information.
Check your understandingDoes getting a random result prove the device made an error?
No. First ask what the ideal model predicts. Randomness can be part of the intended experiment.
Sources: noise and errors · an approach to fault tolerance
Read a headline with better questions
Ask what task was solved, how correctness was checked, which classical method was compared and which costs were counted. A useful result can still have a narrow scope. A small demonstration, an architectural proposal and a practical advantage are different achievements.
A concrete example
Synthetic headline: “Quantum is faster.” Before accepting that conclusion, find the task, accuracy, classical comparison and full resource costs.
Keep in mind: Qubit count alone cannot establish that one machine is better at a particular task.
Check your understandingWhat should you ask before accepting a speedup claim?
Faster at which task, compared with what, at which accuracy and with which costs included?
Sources: A Framework for Quantum Advantage
Take the next step.
Sources and boundaries.
Original introductory explanations supported by the references below. Sources checked 24 October 2026. Examples are teaching models; they do not report new hardware results or establish a practical quantum advantage.
- IBM Quantum Learning: states, measurement and single-qubit operations
Amplitudes, measurement probabilities and the H and Z operations.
- IBM Quantum Learning: limitations on quantum information
Relative and global phase; the distinction between the plus and minus states.
- IBM Quantum Learning: quantum mechanics basics
Gate action, circuit notation and the Bell-state example.
- IBM Quantum Learning: multiple quantum systems
Entangled states, joint measurements and why a distant measurement cannot change local outcome probabilities.
- IBM Quantum Learning: circuits
Circuits as ordered operations on quantum and classical information.
- IBM Quantum Learning: noise and errors
Unwanted interactions, control imperfections and error models.
- IBM Quantum Learning: an approach to fault tolerance
Protecting encoded information and implementing reliable operations.
- Lanes and colleagues: A Framework for Quantum Advantage
Defining and validating quantum advantage against classical methods; version 2, 14 July 2025.