Starting with ordinary bits
A bit is the smallest unit of information in a conventional computer. It is either 0 or 1. Physically, those two values might be represented by a low or high voltage, a magnetized or unmagnetized spot, or a charged or uncharged cell. Whatever the physical form, a bit is always in one definite state that can be read without changing it. Text, photos and programs are all long strings of such bits.
What a qubit is
A qubit, short for quantum bit, is the basic unit of information in a quantum computer. It is built from a physical system that follows quantum mechanics, and which has two distinguishable states that are labeled 0 and 1. Various physical systems have been used for this in experiments, including tiny superconducting circuits, individual trapped atoms or ions, and particles of light. Each approach has different strengths and weaknesses, and no single one has yet become the clear standard.
What makes a qubit different from a bit is that, before it is measured, it can be in a combination of its two states, known as a superposition.
Superposition, without the myths
A popular but misleading summary says a qubit “is both 0 and 1 at the same time”. A more careful description is that the qubit’s state is described by two numbers, called amplitudes, one tied to 0 and one tied to 1. These amplitudes are related to the probabilities of what you will see when you measure the qubit. If the amplitudes favor 0, you will usually get 0 but sometimes 1; if they are balanced, the two outcomes are about equally likely.
The key facts are:
- A measurement always gives a plain 0 or 1, never a blend.
- The measurement changes the qubit. After it, the qubit is in the state that was observed, and the earlier amplitudes are gone.
- The amplitudes can carry a phase, a kind of direction, that does not show up as a probability by itself but affects how amplitudes combine.
That last point matters. Amplitudes can add up or cancel, similar to waves, which is called interference. Quantum algorithms are designed so that the wrong answers tend to cancel out and the right answers reinforce. This is a more accurate picture than the claim that a quantum computer “tries every answer at once and picks the best”, which overstates things. A computer cannot read out all the possibilities; it gets a single measured result, so an algorithm must be arranged so that the result it reads is likely to be useful.
How the numbers grow
The strange power of quantum states comes when several qubits are put together. Describing the state of 1 qubit takes 2 amplitudes. For 2 qubits, there are 4 possible outcomes (00, 01, 10, 11), so 4 amplitudes. For 3 qubits it takes 8, and in general, for n qubits, 2 to the power n amplitudes. That means 10 qubits need 1,024 amplitudes, and 20 qubits need 1,048,576.
Ordinary bits do not grow this way: 3 bits are in just one of 8 possible states at any moment, and you need only 3 values to record which one. For qubits, the full description of the quantum state needs all 8 amplitudes. This rapid growth is why simulating even a modest quantum computer on a normal computer becomes hard as the qubit count rises, and it is one reason researchers think quantum machines may help with some problems.
It does not mean a quantum computer is faster at everything. For many everyday tasks, a conventional computer is better and cheaper. The advantage is expected only for specific kinds of problems, such as simulating quantum systems in chemistry and materials, and certain mathematical tasks. The broader picture is covered in the discussion of quantum computing.
Entanglement
Qubits can also be entangled, meaning their joint state cannot be described as separate states for each qubit. Measurement results on entangled qubits are correlated in ways that have no everyday counterpart. Entanglement is a resource that quantum algorithms use, but it does not let information travel faster than light, and it is frequently misdescribed in popular accounts.
Why qubits are hard to build
- Fragility. Qubits are easily disturbed by heat, vibrations, stray electromagnetic fields and even the surrounding material. This disturbance, called decoherence, destroys the quantum behavior. Many systems operate at extremely low temperatures or in vacuum chambers to limit it.
- Errors. Operations on qubits are imperfect, and error rates are much higher than in ordinary electronics, where a bit flip is rare. Ordinary bits can be protected by simple repetition. Quantum states cannot be copied freely, so quantum error correction uses cleverer schemes that spread one logical qubit over many physical qubits. Estimates of how many physical qubits are needed per reliable one vary and depend on the hardware and error rate, and they can be large.
- Scale. Controlling and connecting large numbers of qubits is a major engineering challenge.
For this reason, researchers distinguish between the number of physical qubits and the number of reliable, error-corrected logical qubits a machine can offer. Counting qubits alone says little about how useful a machine is.
Bits and qubits compared
| Feature | Bit | Qubit |
|---|---|---|
| Possible readout | 0 or 1 | 0 or 1 |
| State before reading | Definite 0 or 1 | Described by amplitudes for both |
| Effect of reading | None | Changes the state |
| Copying | Easy | Not possible for an unknown state |
| Error behavior | Rare, easily corrected | Frequent, hard to correct |
| Typical hardware | Mass-produced chips | Experimental devices |
A sensible takeaway
Qubits extend the idea of a bit by allowing amplitudes, interference and entanglement, while still giving ordinary answers when measured. They are a real and well-understood part of physics, but building large, dependable machines is still a research effort, and predictions about when useful ones will arrive remain uncertain.