The Road Ahead: Evaluating Claims Yourself

The road from today's noisy devices to fault-tolerant quantum computing is real, but it has no timetable. Error correction is demonstrated at small scale. Useful error-corrected computing is not. What you take from this course is a lasting skill. It is a filter that sorts claims into proven theory, lab demonstrations, practical ability, and roadmap claims. You also have the arithmetic habits to apply it to any headline, without trusting anyone — including us.

Where are we on the road from NISQ to fault tolerance?

NISQ stands for noisy intermediate-scale quantum. John Preskill coined the term in 2018. It describes today's era: devices with tens to about 1,000 physical qubits. Their gates fail roughly once every 100 to 1,000 operations. No error correction protects the computation. Every result in this course's honest-evidence sections is a NISQ result. That includes small-molecule chemistry, QAOA experiments, and sampling demonstrations. They are real physics at small scale, with no proven business value yet.

The goal is fault tolerance. That means encoding each logical qubit (a reliable, error-corrected qubit) across many noisy physical qubits. Errors are then found and fixed faster than they pile up. This has been proven theory since the "threshold theorems" of the 1990s.

Where does the road actually stand? Experiments have now shown small numbers of logical qubits whose error rates get better as the code grows. That is the key sign that error correction works as the theory says. It is demonstrated at small scale. What does not exist anywhere is a machine running deep algorithms on many logical qubits. That is the level chapter 6's RSA arithmetic needs: millions of physical qubits, versus today's roughly 10³ (see qubit count). The gap is measured in orders of magnitude, not in months. See our fault-tolerance reality page for the sourced timeline of what has actually been shown.

Think of learning to fly. Gliding a few meters off a hill is real flight. But it is a long way from a plane that carries passengers across an ocean. Where the picture breaks: early fliers could see the ocean they wanted to cross. With quantum computing, nobody knows exactly how far away the "other shore" is.

Between here and there lies "early fault tolerance." These would be machines with a handful to a few hundred logical qubits. Which uses will those first machines unlock? Quantum simulation is the strongest candidate, as in chapter 2. This is the biggest open question in the field. Anyone who answers it with confidence is selling something.