153 | John Preskill on Quantum Computers and What They're Good For

Episode Summary

Executive Summary: Sean Carroll and John Preskill discuss quantum computing’s current state, why Preskill shifted into the field, and what quantum machines are likely to be good for: simulating quantum systems, certain cryptographic tasks, and eventually error-corrected large-scale computation. They emphasize both genuine promise and the danger of hype, while also connecting quantum information to black holes, holography, and emergent spacetime.

Main Topics: Preskill’s career shift into quantum information (Priority: 5/5): Preskill explains how the SSC cancellation, curiosity about black holes and information, and Shor’s factoring algorithm led him from particle physics into quantum computing. What a quantum computer is and why entanglement matters (Priority: 5/5): The conversation explains qubits, superposition, measurement limits, and entanglement as the key resource that makes quantum information fundamentally different from classical information. Likely applications and limits of quantum computing (Priority: 5/5): Preskill argues the strongest near-term case is simulating quantum systems in chemistry/materials, while remaining skeptical about broad claims for optimization and everyday consumer use. Hardware platforms and engineering challenges (Priority: 4/5): They compare trapped ions, superconducting circuits, and newer approaches like Rydberg atoms, focusing on coherence, readout, and especially two-qubit gate error rates. Quantum error correction as a scientific breakthrough (Priority: 5/5): Error correction is presented not merely as engineering, but as the conceptual method that lets large entangled systems remain quantum despite decoherence. Quantum information, black holes, and emergent spacetime (Priority: 4/5): The discussion ends with how quantum error correction and holography illuminate black holes and suggest spacetime may emerge from entanglement.

Key Arguments: Preskill’s move from particle physics to quantum computing was driven by both external events (SSC cancellation) and intellectual interest in quantum information and black hole information. Quantum computers are not just faster classical computers; they process information in a fundamentally different way enabled by entanglement. The vast classical description of an entangled quantum state does not automatically translate into computational usefulness; algorithms must exploit structure and measurement constraints. The most credible long-term application of quantum computing is simulating strongly entangled quantum systems in chemistry and materials science. Shor’s algorithm showed that quantum computers can solve some classically hard problems very efficiently, especially factoring, with major implications for cryptography. Near-term noisy quantum devices are useful for experiments and for learning how to build better, error-corrected machines, even if they are not yet commercially transformative. Quantum error correction works by encoding information in highly entangled states so the environment cannot easily extract or destroy it. There is no strong theoretical basis to expect dramatic quantum speedups for all optimization problems; claims in that direction should be treated cautiously. Holographic duality and quantum error correction provide a concrete framework in which spacetime geometry appears to emerge from entanglement. The field is advancing, but progress will require sustained investment and realistic expectations rather than hype.

Data Points: Shift to quantum information: ~25 years ago - Preskill says the transition from particle physics to quantum computing occurred about 25 years earlier, in the mid-1990s. SSC cancellation time frame: mid-1990s - The cancellation of the Superconducting Super Collider helped motivate Preskill to explore a new research direction. Shor’s factoring algorithm: 1994 - Preskill identifies the discovery as a major turning point that made quantum computing seem serious and consequential. Teaching-based transition: 1997 - He says teaching a year-long course on quantum information helped him synthesize the field and deepen his understanding. Feynman’s proposal: May 1981 - Preskill cites Feynman’s proposal that quantum computers should simulate quantum systems. Current device scale: 50 to 100 qubits - He describes near-term quantum devices as being in the 50-100 qubit range. Google Sycamore qubits: 53 working qubits - He uses Google’s 2019 Sycamore system as an example of a noisy but classically hard-to-simulate quantum device. Sycamore depth: up to 20 layers - He notes the 2019 experiment ran circuits with about 20 layers of entangling gates. Sycamore output reliability: about 1 time out of 500 - He says the right answer was obtained roughly once per 500 runs due to noise. Classical simulation time: at least days - He compares Sycamore’s task to classical supercomputer simulation that can take days. Two-qubit gate error rate: about 1% - He characterizes current best hardware as still having around one-percent error in entangling operations. Desired gate error rate: 10^-5 - He argues hardware needs to improve to around ten-to-the-minus-five error rates for scalable error correction. Ion-trap scaling limit: more than 32-64 ions becomes difficult - He says trapped-ion systems become hard to scale beyond roughly 32 to 64 ions because of mode complexity. RSA factorization target: 2,000-bit numbers - He explains that breaking widely used RSA would require factoring numbers around 2,000 bits long. Exhaustive search speedup: square-root improvement - He states quantum computers can speed up exhaustive search quadratically, not exponentially.

Pivotal Quotes: "It’s not just a computer like the ones we have now, but much, much faster. It really processes information in a very different way." — John Preskill: Defining what makes a quantum computer fundamentally distinct from a classical one. "We can outsmart decoherence." — John Preskill: Summarizing the core idea behind quantum error correction and scalable quantum computation. "The best idea we have is simulating quantum systems with quantum computers." — John Preskill: Explaining why quantum simulation is the most credible long-term application of quantum computing.

Implications: Quantum computing is real but still immature: the main near-term value is scientific experimentation and technology development, while the biggest practical payoffs likely lie in quantum simulation, cryptography changes, and future error-corrected machines.