Why Quantum Computing Could Change Everything

Thesis: Quantum computing is real science wrapped in unreliable marketing. Error correction is the entire problem, and most predicted applications will never materialize.

Quantum computing is described as a technology that will break encryption, design new drugs, and solve problems classical machines cannot touch. It is also, at present, a field where laboratory machines struggle to outperform a laptop on useful tasks. I wanted to understand that contradiction. What I concluded is that the underlying physics and engineering are genuinely extraordinary, while the public narrative has been shaped by funding incentives that reward exaggeration. Separating the two seems worth doing, because misplaced expectations eventually produce disillusionment that damages good research.

The core technology

A classical bit is zero or one. A qubit occupies a superposition of both, and multiple qubits can be entangled so that their states cannot be described independently. An algorithm manipulates amplitudes so that interference cancels wrong answers and reinforces right ones. Physical implementations vary: superconducting circuits cooled to near absolute zero and controlled by microwave pulses; trapped ions manipulated by lasers; neutral atoms held in optical tweezers; and photonic approaches. All share a fundamental fragility: qubits decohere within microseconds to seconds as they interact with their environment, and every gate operation introduces error. Measurement also destroys superposition, so algorithms must be arranged so that the single reading taken at the end contains the answer.

Figure 1. A superconducting quantum processor suspended inside a dilution refrigerator, cooled to near absolute zero — the physical setup needed just to keep qubits coherent long enough to compute.

Recent developments and real problems

The most significant recent achievement is experimental evidence that quantum error correction can work below threshold, meaning that adding more physical qubits to encode one logical qubit reduces the error rate rather than increasing it. This is the necessary precondition for everything else. The scale required, however, is sobering: useful cryptographically relevant computation is generally estimated to need millions of physical qubits, against machines that currently operate in the hundreds to low thousands. Meanwhile many advertised applications, particularly in optimization and machine learning, lack any proven asymptotic advantage over good classical algorithms, and several claimed advantages have been overturned by improved classical methods. Quantum advantage claims have a poor track record for this reason: a demonstration only means something against the best classical algorithm available, and that keeps improving in response.

Figure 2. Logical error probability versus surface-code size on Google's Willow chip. Larger qubit grids (5x5, 7x7) show falling error rates — the 'below threshold' result central to this essay's argument.

Outlook and solutions

I expect the credible applications to be narrow and chemical: simulating molecular and material systems whose quantum behavior is intrinsically hard to represent classically. That alone would justify the investment. The nearer-term obligation is defensive, migrating to post-quantum cryptography now, because data intercepted today can be stored and decrypted later. The engineering priorities are cryogenic control electronics that can address thousands of qubits without a wire per qubit, better materials to extend coherence, and honest benchmarking standards that report error rates and problem classes rather than raw qubit counts. Cryogenic infrastructure is a serious constraint in itself, since dilution refrigerators consume substantial power and offer limited space for the control wiring each additional qubit demands.

Figure 3. Racks of control electronics wired into a dilution refrigerator. Every additional qubit needs its own control line, which is exactly the wiring bottleneck this essay flags as an engineering priority.

Conclusion

My personal insight is about how to evaluate emerging technology. When a field cannot state clearly what problem it will solve, by when, and how the claim would be falsified, the appropriate response is patient skepticism rather than either enthusiasm or dismissal. Quantum computing is worth funding as fundamental research on materials and control engineering. It is not worth planning a career around a revolution scheduled by press release.

Sources

1. Google Quantum AI and Collaborators (2025). Quantum error correction below the surface code threshold. Nature 638, 920–926. https://www.nature.com/articles/s41586-024-08449-y

2. Preprint version with full error-rate data (Λ = 2.14; 0.143% per cycle; rare correlated errors roughly hourly). arXiv:2408.13687. https://arxiv.org/pdf/2408.13687