In a quantum computer, a qubit is like a tightrope walker: any puff of wind and balance is lost. Errors accumulate, and the computation falls apart. Usually, they are corrected by comparing neighboring qubits, but that's like asking the tightrope walker to glance at his neighbors — extra movements and risk. In the new work, physicists did it differently: the qubit, when it stumbles, cries out 'I fell!' and is immediately caught.

For this, they used a superconducting transmon — a qubit with three energy levels, like a staircase with three steps. Logical zero and one are the bottom and top steps, and the middle one is a trap for errors. If the qubit accidentally slips, it doesn't land in a wrong state but on the middle step, and that's immediately visible. A special auxiliary qubit, the observer, checks whether the main qubit has stumbled and, if necessary, puts it back in place.

The scientists also taught the auxiliary qubit to do two jobs at once: watch for errors and check parity when creating entangled pairs. It's like one assistant who both spots the tightrope walker and hands him the pole. As a result, they obtained pairs of qubits linked by quantum entanglement, with a warning about any error. This approach doesn't require new complex circuits: it works on existing quantum chips.

Quantum computers promise to solve problems beyond the reach of classical ones — from simulating molecules to optimizing routes. But without reliable error correction, they remain just expensive toys. The scheme with 'erasure' qubits brings closer the day when a quantum processor will compute for a long time without glitches, like a tightrope walker who has learned to fall safely.