What a qubit actually buys you

Quantum computing gets explained in one of two ways: a cat that is somehow alive and dead, or a machine that “tries every possible answer at the same time.” The first is a physics riddle that helps nobody, and the second is wrong in a way that matters. Here is the version that actually holds up.

A classical bit is a switch: 0 or 1. A qubit is not a switch stuck halfway. It is a thing whose measurement will come out 0 or 1 with some probability — and, crucially, that probability is carried by numbers that can be positive or negative. That sign is the whole trick. Nothing else about quantum computing makes sense until you have it.

A qubit in superposition is not secretly holding both answers. It is in a state that, when you look, resolves to one of them. Put a hundred qubits together and you can describe a state spread across an astronomical number of combinations — but you never get to read all of them. Measure, and the whole thing collapses to one plain answer, chosen by those probabilities.

So if you could only prepare a big superposition and measure it, you would have an extremely expensive random-number generator. The magic is not the spreading out. It is what you can do to the state before you measure.

Because those probability amplitudes have signs, they can cancel. A quantum algorithm is a carefully choreographed sequence of operations designed so that the paths leading to wrong answers cancel out — negative meets positive — while the paths leading to right answers add up. When you finally measure, the answer you want is the one left standing.

That is the honest one-sentence definition of a quantum speedup: arrange interference so the wrong answers destroy each other. It is closer to noise-cancelling headphones than to a search that checks everything. And it is why writing quantum algorithms is hard — you are composing an interference pattern, not writing a loop.

Only some problems have a structure that interference can exploit, so the list of real speedups is short. Factoring large numbers (Shor’s algorithm), which is why quantum computing worries cryptographers. Simulating quantum systems themselves — molecules, materials, reactions — which is arguably the point, since that is a job classical machines are genuinely bad at. And a modest, provable speedup for brute-force search (Grover’s algorithm): the square root of the work, useful but not the exponential leap people imagine.

For most of what a computer does — spreadsheets, databases, rendering, the model behind your chatbot — a quantum computer offers nothing. It is not a faster classical computer. It is a different instrument that is extraordinary at a few things and irrelevant to the rest.

Drop “tries everything at once” and keep this instead: a qubit is a probability with a sign, and a quantum algorithm steers those signs so wrong answers cancel. When a headline promises a quantum computer will revolutionise some everyday task, the useful question is simply — does that problem have the kind of structure interference can act on? Usually the answer is no, and that is not a disappointment. It is the shape of the technology.