What is quantum computing, and what can it actually do?
A fundamentally different model of computation using qubits rather than bits — and its realistic near-term applications are far narrower than the coverage suggests. It is not a faster computer; it is a machine that can do a small number of specific things in a way classical computers cannot.
What a qubit is. A classical bit is 0 or 1. A qubit can be in a superposition of both, described by amplitudes. Crucially, qubits can be entangled, so their states are correlated in ways with no classical equivalent. With n entangled qubits the system's state requires 2ⁿ numbers to describe — which is where the power comes from, and also why simulating a large quantum computer classically is infeasible.
The persistent misconception. Quantum computers do not try all answers simultaneously and pick the right one. Measuring a superposition returns one outcome at random. A useful quantum algorithm must arrange interference so that wrong answers cancel out and the right one becomes probable — which only a handful of known algorithms achieve.
Where the advantage is real:
Factoring and discrete logarithms, via Shor's algorithm, which would break the public key cryptography currently securing the internet. This is the driver behind post-quantum cryptography standards being deployed now, and behind "harvest now, decrypt later" concerns.
Simulating quantum systems — molecules, catalysts, materials, superconductors. This is the most credible commercial application, because nature is quantum and classical simulation scales terribly.
Unstructured search, via Grover's algorithm, giving a quadratic rather than exponential speedup — real but modest.
Optimisation and machine learning, where claims substantially outrun demonstrated results.
Why it is hard. Qubits lose their state through decoherence in microseconds, and error rates are high. Error correction requires many physical qubits to make one reliable logical qubit — plausibly thousands. Most systems need near-absolute-zero temperatures and extreme isolation.
Where things stand. Machines with hundreds to thousands of noisy qubits exist and logical-qubit demonstrations are progressing. Cryptographically relevant machines do not yet exist, and timelines remain genuinely uncertain.