Every introduction to quantum computing makes the same mistake.
"A qubit can be 0 and 1 at the same time." The sentence is everywhere. It is short, it moves fast, and it gives you just enough intuition to feel like you understand something. But it encodes a fundamentally wrong picture of what superposition is, and that wrong picture compounds into bigger misunderstandings every step after it.
A qubit in superposition is not "both 0 and 1." It is a single mathematical object in a state that cannot be reduced to either. That distinction sounds subtle. It is not.
Here is what is actually happening, in five steps.
1. The State Has No Value. It Has a Vector.
A classical bit is binary and definite. At every moment, it is 0 or 1. A transistor conducts current or it does not. The bit always has a value.
A qubit in superposition has no value. What it has is a state:
|ψ⟩ = α|0⟩ + β|1⟩
Where α and β are complex numbers called probability amplitudes, constrained by the normalization condition:
|α|² + |β|² = 1
This is not describing two simultaneous values. It is describing one vector in a two-dimensional complex Hilbert space. The |ψ⟩ is its own mathematical entity, distinct from both |0⟩ and |1⟩.
Three things follow immediately from this:
- Superposition does not mean the qubit "is both." |ψ⟩ is not a blend of |0⟩ and |1⟩ in the way gray is a blend of black and white. It is a different kind of object entirely.
- α and β are complex numbers, not percentages. Their squared moduli give measurement probabilities, but the amplitudes themselves carry phase information that determines interference, and interference is where the computation lives.
- When you measure, the state collapses. You get |0⟩ with probability |α|², or |1⟩ with probability |β|². Before that moment, neither outcome existed.
2. The Bloch Sphere: Every Point Is a Possible Quantum State
There is a clean geometric way to visualize all possible pure states of a single qubit. It is called the Bloch sphere.
Any pure qubit state can be written using two real parameters, θ and φ:
|ψ⟩ = cos(θ/2)|0⟩ + eiφ·sin(θ/2)|1⟩
These two angles map to a unique point on the surface of a unit sphere:
- North pole (θ = 0): the state |0⟩. Measuring gives 0 with 100% certainty.
- South pole (θ = π): the state |1⟩. Measuring gives 1 with 100% certainty.
- Equator (θ = π/2): maximum superposition, 50% probability for each outcome.
But the Bloch sphere shows something the probability numbers alone do not. The angle φ is the relative phase. It does not change measurement probabilities at all. Two states at the equator with φ = 0 and φ = π both give 50/50 outcomes, every single measurement. Yet they are different states, and they behave completely differently inside an algorithm, because phase determines interference.
A state that looks identical from the outside can produce completely different computational outcomes. This is not a quirk. It is the mechanism.
3. The Double Slit: Where Superposition Becomes Visible
The cleanest physical demonstration of superposition and interference is the double-slit experiment.
Fire a particle at a barrier with two slits, A and B. Without any measurement of which slit the particle passes through, the particle travels as a linear combination of both paths:
|ψ⟩ = 1/√2 · |A⟩ + 1/√2 · |B⟩
The amplitudes of both paths interfere on the screen. Where they add constructively, you get bright bands. Where they cancel destructively, you get nothing. The characteristic alternating pattern appears.
Now ask which slit the particle went through. The moment you measure the path, the superposition collapses. The particle is in one slit. The interference disappears. Two bands remain, exactly where classical physics would predict them to be.
The particle did not choose a slit and we missed it. The superposition was real, and the measurement ended it.
Tonomura et al. showed in 1989 that this pattern builds up one particle at a time. Each individual particle, traveling in superposition, contributes to the interference. The pattern is not a statistical accident from many particles piling up. It is a property of the quantum state itself.
This is why measurement changes outcomes. Not because the measurement device is poorly calibrated. Because measuring which slit the particle used is precisely what destroys the superposition that was producing the interference.
4. Measurement Destroys Superposition. It Does Not Reveal It.
This is where the coin-under-a-cup analogy breaks, permanently.
The coin analogy suggests the qubit already has a value, and measurement simply reveals it. The uncertainty is yours, not the system's.
Quantum mechanics says something else. The state |ψ⟩ = α|0⟩ + β|1⟩ has no hidden value. It is the complete description of the system. When you measure, the quantum system interacts with the measuring apparatus. The environment entangles with the system. The state collapses irreversibly, probabilistically, to one of the eigenstates.
The result was not predetermined. The act of measuring creates it.
This is not a matter of philosophical preference. The Bell inequalities, proven in 1964 and confirmed experimentally starting in the 1970s, ruled out local hidden variable theories. No pre-existing hidden value can account for the correlations produced by quantum systems. The measurement outcome is genuinely indeterminate before measurement.
What follows:
- Each measurement of the same |ψ⟩ can give a different result. This is not imprecision in the equipment. It is fundamental.
- Superposition is the real state of the system. There is no hidden fact underneath it waiting to be discovered.
- The measurement postulate: after obtaining outcome m, the post-measurement state is the eigenvector of the measured observable associated with m.
The superposition is destroyed by measurement, not decoded.
5. This Is Why Quantum Computing Is Different
Here is where the linear combination stops being abstract.
N qubits in superposition represent not one state but 2^N amplitudes simultaneously. A single state vector of N qubits encodes a complex coefficient for every possible N-bit string.
- 3 qubits: 2³ = 8 basis states → 8 complex amplitudes
- 10 qubits: 2¹⁰ = 1,024 basis states → 1,024 amplitudes
- 50 qubits: 2⁵⁰ basis states → ~1.1 × 10¹⁵ amplitudes
- 300 qubits: 2³⁰⁰ basis states → more than atoms in the observable universe
A classical 3-bit register holds exactly one value. A quantum register of 3 qubits holds a weighted superposition across all 8 possible values, simultaneously, in a single state.
Quantum gates are unitary operations. They act on the entire amplitude vector at once. The algorithm does not run 2^N separate computations in parallel. It manipulates the full amplitude structure through carefully chosen transformations.
The useful analog: a classical die falls on one number. A quantum die exists in superposition of all its outcomes, each carrying an amplitude. The algorithm configures those amplitudes so that the correct result accumulates the highest probability before measurement.
But you cannot read out all 2^N amplitudes. When you measure, you get one outcome, sampled from the probability distribution the algorithm has engineered. The design problem is interference: suppressing wrong answers, amplifying right ones. Grover's algorithm does this for unstructured search. Shor's does it for factorization. Both are interference structures, not classical computations running fast.
This is not a faster classical computer. It is a structurally different computational model operating in a space classical hardware cannot replicate.
What the Framing Gets Wrong, and Why It Matters
"Being 0 and 1 at the same time" is not a harmless simplification. It leads directly to two persistent mistakes.
The first: expecting quantum computers to solve everything faster by trying all options simultaneously. They do not. Speedup is problem-specific and requires algorithm structures that do not exist for most computations.
The second: assuming the engineering challenge is just getting more qubits. It is not. Without coherence, without precise amplitude control, without error correction, more qubits means more noise. The qubit in superposition is fragile, and any unwanted interaction with the environment is effectively a measurement that collapses it.
The next piece in this series is quantum error correction, which is where the real engineering challenge begins. That challenge starts here, with understanding what superposition actually is, and why maintaining it is so hard.
This is part of an ongoing series on quantum computing from the ground up. Previous entries: The End of the Transistor Era · One Story, Two Qubits: Entanglement Explained
Sources
- Nielsen, M.A. & Chuang, I.L., Quantum Computation and Quantum Information, Cambridge University Press (2000)
- Bloch, F., Nuclear Induction, Physical Review 70 (1946)
- Feynman, R., The Feynman Lectures on Physics, Vol. 3, Addison-Wesley (1965)
- Tonomura, A. et al., Demonstration of single-electron buildup of an interference pattern, Am. J. Phys. 57 (1989)
- Dirac, P.A.M., The Principles of Quantum Mechanics, Oxford University Press (1930)
- Preskill, J., Quantum Computing in the NISQ era and beyond (2018)