What a Qubit Actually Is, and Why Fixing Its Errors Is the Hard Part
Building a qubit is difficult enough, but keeping it reliable long enough to compute anything useful is the problem that has occupied physicists for decades.
Start with the coin, not the computer
A classical computer bit is like a coin lying flat on a table: heads or tails, 0 or 1, always one or the other. A qubit is more like that same coin spinning in the air. While it spins, it is not simply heads or tails, it holds a combination of both possibilities at once, described by physicists as a superposition. Only when the coin lands, or the qubit is measured, does it settle into a definite 0 or 1.
This spinning state is where quantum computing gets its potential power. Link several qubits together so their spins depend on each other, a property called entanglement, and you can represent and manipulate a huge number of possibilities simultaneously. Certain problems, such as simulating how molecules behave or searching through vast combinations, could in theory be tackled far faster than any ordinary computer could manage.
Why qubits are so easily disturbed
The catch is that the spinning coin is extraordinarily delicate. A qubit can be built from many different physical things: a trapped ion, a loop of superconducting metal cooled close to absolute zero, a single electron’s spin, a photon of light. Whatever the substrate, the same problem applies. The delicate superposition only survives if the qubit is almost completely isolated from its surroundings.
In practice, total isolation is impossible. Stray heat, tiny vibrations, electrical noise, even cosmic rays passing through the lab, all nudge the qubit. This unwanted interaction is called decoherence, and it causes the spinning coin to collapse into a definite 0 or 1 before the calculation is finished, or to drift into the wrong state entirely. The timescale over which a qubit stays reliably usable, its coherence time, is measured in fractions of a second for most current technologies. Every operation performed on a qubit also has some chance of introducing a small error, simply because the control signals themselves are imperfect.
Why you cannot just copy the answer and check it
Classical computers deal with errors constantly, and the fix is usually simple: store the same bit in three places and take a majority vote. If one copy flips by accident, the other two outvote it.
This trick does not work for qubits, for two deep reasons. First, a rule of quantum mechanics called the no-cloning theorem says it is impossible to make an exact copy of an unknown quantum state. You cannot simply duplicate a qubit the way you duplicate a bit. Second, directly measuring a qubit to check its value destroys the very superposition that makes it useful, collapsing the spinning coin the moment you look at it.
So error correction has to work without ever directly reading the information stored in the qubit. This is the central puzzle that has made quantum error correction one of the hardest problems in applied physics.
How quantum error correction actually works
The solution physicists have developed is to spread one unit of quantum information across many physical qubits, forming what is called a logical qubit. Extra qubits are woven in purely to act as sensors. By carefully measuring relationships between qubits, rather than the qubits themselves, it is possible to detect that an error has occurred and roughly what kind it was, without ever learning or disturbing the actual information being protected.
This is genuinely difficult to engineer. Detecting and correcting errors requires extra qubits and extra circuitry, and every one of those additional components can itself introduce new errors. There is a threshold effect: if the error rate of the underlying physical qubits is low enough, adding more of them to build a logical qubit makes the overall system more reliable. But if the physical error rate is too high, adding more qubits simply adds more ways for things to go wrong. Getting physical qubits reliable enough to sit on the right side of that threshold, consistently and at scale, has taken the field decades.
The practical consequence is a steep overhead. Building one highly reliable logical qubit can require dozens or even hundreds of physical qubits, all of which must be controlled, cooled, and read out in a coordinated way. This is the main reason that headlines about a chip with a large number of qubits do not translate directly into a machine that can run useful, long calculations. A processor with many noisy physical qubits may still only yield a handful of dependable logical qubits, if any.
Why this matters for what comes next
Understanding this distinction, between raw physical qubits and error-corrected logical qubits, is the key to reading quantum computing news sensibly. Progress is genuinely being made on coherence times, error rates and the engineering of error correction codes, but it is incremental and hard-won, not a switch that suddenly flips from experimental to world-changing. Anyone assessing claims in this field should ask not just how many qubits a device has, but how reliable those qubits are and what fraction of them are needed just to keep the others honest.
For readers wanting to track genuine progress rather than hype, official research funding bodies and standards organisations publish regular, sober assessments of where the technology actually stands, which is a more reliable guide than any single product announcement.