Superposition without the hype
On this page
In Part 1 we said a qubit is a blend of 0 and 1 until you measure it. That blend has a name, superposition, and it usually gets described in breathless terms: "in two places at once", "every possibility simultaneously", "parallel universes". This piece takes the mystery out. You'll need a few small numbers and nothing you'd want a calculator for.
The blend is just two numbers
A qubit's state is described by two numbers called amplitudes: one attached to 0 and one attached to 1. To get the chance of each result, you square its amplitude. (Strictly, you square its size; amplitudes also carry a phase, which we'll come to.)
Take an even blend. Each amplitude is about 0.71. Square it: 0.71 × 0.71 ≈ 0.5. So there's a 50% chance of reading 0 and a 50% chance of reading 1. The two chances always add up to 100%, because the qubit has to give you something when you look.
Change the amplitudes to 0.6 and 0.8 and the odds shift. 0.6 × 0.6 = 0.36 and 0.8 × 0.8 = 0.64, so now you'd read 1 about 64% of the time. Set them to 1 and 0 and you'd read 0 every single time, which is just an ordinary bit. That's what the slider in the home page demo is doing: moving the amplitudes and showing you the odds that result.
Picture a globe
Physicists often draw a qubit's state as a point on the surface of a globe. Put 0 at the North Pole and 1 at the South Pole.
A qubit sitting exactly at the North Pole is a plain 0. At the South Pole it's a plain 1. Anywhere else is a superposition. The closer the point is to the North Pole, the more likely you are to read 0. Points on the equator are even 50/50 blends.
Notice something: there isn't just one point on the equator, there's a whole ring of them. Every point on that ring gives the same 50/50 odds if you measure straight away, yet they are different states. The difference between them is called phase, and it's like longitude on the globe. Phase doesn't change the odds on its own, but it changes how a qubit behaves when it's combined with others or nudged by the computer. A quantum computer's operations are essentially rotations of this globe.
How likely each answer is depends on latitude. How the qubit will interfere depends on longitude. Quantum algorithms use both.
The part that matters: amplitudes can be negative
If amplitudes were just probabilities in disguise, quantum computers would be no more powerful than flipping weighted coins. The difference is that an amplitude can be negative. (In full generality it can be stranger than that, which is what phase captures, but negative is enough to see the idea.)
Squaring hides the sign, so +0.71 and −0.71 give the same 50% odds if you measure right away. But when a calculation combines amplitudes before you measure, the sign changes everything. Positives and negatives can cancel, just like waves on water.
A tiny worked example
Suppose a calculation offers two routes to the answer 1. Along one route the amplitude is +0.5. Along the other it's −0.5. Add them: +0.5 + (−0.5) = 0. Square zero and you get zero. The answer 1 never shows up.
Meanwhile two routes lead to the answer 0, each with amplitude +0.5. Add them: 0.5 + 0.5 = 1. Square it: 1, or 100%. You get 0 every time.
That isn't a made-up toy. There's a basic quantum operation, the first one every course teaches, that takes a plain 0 and turns it into an even blend. Apply it once and you get 50/50 odds. Apply it a second time and the routes interfere exactly as above: the 1 cancels out and you're back to a certain 0. The blend in the middle was real, and the cancelling is what brought it back.
Now scale that up. A quantum algorithm is a long sequence of operations designed so that, across thousands or millions of possible answers, the routes to wrong answers cancel and the routes to the right answer pile up. When you finally measure, you're likely to land on the right one.
Quantum algorithms are recipes for arranging cancellations: wrong answers interfere away, right answers build up.
How a real algorithm uses superposition
Most quantum algorithms follow the same broad shape, and superposition is the first move.
Step one: spread out. Every qubit starts as a plain 0. The computer then puts each one into an even blend. With three qubits, that gives eight possible results, from 000 to 111, each with the same small amplitude. With fifty qubits it's about a thousand million million, all held at once in one physical system. Nothing useful has happened yet; the computer has simply laid out every possibility on the table.
Step two: mark and mix. Next comes a sequence of operations tailored to the problem. Some of them flip the sign of particular amplitudes, the ones connected to the answer you want, while others mix amplitudes together so they interfere. This is where the problem's structure matters. The operations have to be designed so that, round after round, amplitude drains away from wrong answers and collects on right ones.
Step three: measure. Finally the qubits are measured. If the design worked, the right answer now carries most of the amplitude, so it's the result you're most likely to see. Researchers usually run the whole thing many times to be confident.
An orchestra tuning up is a fair picture. At first every instrument plays its own note and the hall is full of noise. As the players adjust, the stray notes fade and a single clear chord remains. Superposition provides the noise. Interference provides the tuning.
This shape also explains why quantum computers only help with some problems. If a problem has no structure that lets you design step two, all you get from step three is a random answer from step one.
What about Schrödinger's cat?
You've probably heard of the cat in a box that is supposedly both alive and dead until someone opens the lid. Erwin Schrödinger invented that story in 1935, and he meant it as a criticism. He thought it was absurd to apply superposition to everyday objects like cats.
In practice, large objects don't stay in superposition, because they are constantly bumping into air molecules, light and heat. Each bump acts like a tiny measurement. Qubits only keep their blend because engineers go to extreme lengths to isolate them. So the cat is a thought experiment about where the quantum world ends, not a description of how qubits work.
Why you never see the blend
When you measure, you get one result, and the amplitudes are gone. You can't peek at them mid-calculation without destroying them. That's also why qubits are so delicate. Stray heat, vibration or electrical noise can act like an accidental measurement and smear the blend out before the calculation finishes. Engineers call this decoherence, and fighting it is most of the work in building a quantum computer.
It's also why results come as statistics. Researchers run the same quantum program hundreds or thousands of times, collecting one answer per run, and look at how often each answer appears. That pattern is how they read off what the amplitudes were.
Why more qubits get big fast
One qubit needs two amplitudes. Two qubits need four (for 00, 01, 10 and 11). Three need eight: 000, 001, 010, 011, 100, 101, 110 and 111. Every extra qubit doubles the count.
Ten qubits need 1,024 amplitudes. Twenty need about a million. Fifty need about a thousand million million. By 300 qubits the number of amplitudes is larger than the estimated number of atoms in the observable universe.
That doubling is why ordinary computers struggle to simulate quantum ones in general. The straightforward method stores every amplitude, so each extra qubit doubles the memory it needs, and somewhere around 50 qubits even the largest supercomputers run short. Cleverer methods can do far better when a calculation creates little entanglement or has special structure. That's why every claim of quantum advantage gets tested against them, and why some have been overturned. It's also why the "tries every answer at once" story is tempting. But remember the catch: you can only ever read one result out. The power comes from steering all those amplitudes so the useful result is the one you're likely to read.
Common questions
Does superposition mean parallel universes? Some physicists like that interpretation, called "many worlds", and many don't. You don't need any interpretation to use a quantum computer. The maths makes the same predictions either way.
Is superposition just uncertainty, like not knowing a coin's face? No. If a qubit were simply 0 or 1 and we just didn't know which, the odds from different routes could only add up, never cancel. Interference experiments show outcomes cancelling, which ignorance alone can't produce.
Can I see a superposition? Not directly. Every look is a measurement, and a measurement gives one answer. You infer the superposition from patterns across many runs.
- Superposition is described by amplitudes. Square an amplitude to get the chance of that result.
- Amplitudes can be negative, so they can cancel. That interference is the real source of quantum power.
- Every added qubit doubles the number of amplitudes, but a measurement still gives just one answer.
Sources and further reading
Tags show what kind of source each one is. A standard or government guidance is an official document; a peer-reviewed paper has been checked by other experts; a preprint has not been peer-reviewed yet; an experiment reports a real-world demonstration; a company announcement is the company's own account. Dates and figures were checked against these sources on 11 October 2026. Spotted an error? Email hello@plainquantum.com and it will be corrected, with a note.
- BookQuantum Computation and Quantum Information (10th anniversary edition)Michael Nielsen and Isaac Chuang, Cambridge University Press, 2010
- BookProgramming Quantum ComputersEric Johnston, Nic Harrigan and Mercedes Gimeno-Segovia, O'Reilly, 2019
- CourseBasics of quantum information (free course)IBM Quantum Learning
- ExperimentQuantum supremacy using a programmable superconducting processorArute et al. (Google), Nature, 2019
- Company announcementOn "quantum supremacy"IBM Quantum blog, 2019