Fixing Two Wrong Claims About Future Computers

In the paper's words

Survey literature shapes how funders, program managers, and research groups allocate attention across quantum-computing platforms, so a factual error in a survey propagates into real decisions. This paper repairs two such errors and adds a quantitative scaling analysis. First, it corrects the claim that non-Abelian anyons were experimentally confirmed in 2020 in fractional quantum Hall systems: the 2020 measurements concerned Abelian anyons in the ν = 1/3 state, whose exchange acts by a scalar phase of unit modulus, while for the leading non-Abelian candidate at ν = 5/2 the review literature states that no direct experimental evidence for non-Abelian quasiparticles exists.

The abstract, verbatim; every line below is quoted from the text.

What it claims

level corrections depend on the cited sources' own accounting: [3]'s statement is a review-level assessment, and a direct non-Abelian exchange measurement published after the review's accounting would supersede it.

Numbers it reports

1. The paper retains the constructive program of correlation-first architectures and adds derived numbers: an Abelian exchange modulus $|R| = 1$, a braid length of order $7$ elementary braids per gate at target precision $10^{-3}$, a measurement-iteration count of $100$ for $0.1$ precision in Jones-polynomial estimation, an anyonic error-correction threshold of at least $3.7\%$, and — under explicitly stated assumptions — a fault-tolerance overhead of roughly $10^{3}$ physical qubits per logical qubit, i.e. $10^{5}$–$10^{6}$ physical qubits for $10^{2}$–$10^{3}$ logical qubits.

   from Abstract

2. This paper quantifies that burden under explicitly stated assumptions: with a physical error rate $p = 10^{-3}$ against a code threshold $p_{\mathrm{th}} = 10^{-2}$, each logical qubit costs roughly $10^{3}$ physical qubits, so a commercially useful logical capacity of $10^{2}$–$10^{3}$ logical qubits demands $10^{5}$–$10^{6}$ physical qubits — a million-scale cryogenic control problem.

   from 1. Introduction

3. Explicit derivations, with shown arithmetic, of the quantitative statements the corrected analysis relies on: the unit-modulus scalar exchange phase (Section 4.1), the braid-length scaling estimate at finite target precision (Section 4.2), the measurement-iteration count for Jones-polynomial estimation (Section 4.3), and the anyonic error-correction threshold (Section 4.4).

   from 1. Introduction

4. A physical-qubit overhead derivation under stated assumptions, giving $\sim 10^{3}$ physical qubits per logical qubit and $10^{5}$–$10^{6}$ total for $10^{2}$–$10^{3}$ logical qubits, with uncertainty propagation (Section 4.5).

   from 1. Introduction

5. Second, its error-correction analysis — which reduces the anyon error model to the well-studied toric-code bit-flip model with bit-flip rate $p$ and perfect syndrome measurements, and carries an analytical proof that $p_{c} \geq 3.7\%$ for that problem — supplies the threshold number reported in Section 4.4.

   from 2. Background and Related Work

6. For the error-correction overhead of Section 4.5, we assume a surface-code-style scheme with a physical error rate $p = 10^{-3}$ and a threshold $p_{\mathrm{th}} = 10^{-2}$, and we assume the overhead relation $N_{\mathrm{phys}} \approx \alpha (p_{\mathrm{th}}/p)^{\beta}$ with base factor $\alpha = 10^{2}$ and exponent $\beta = 1$.

   from 3. Methods

How it is done

1. A revised readiness assessment of topological quantum computing that follows from the correction, with falsification conditions, limitations, and a smallest de-risking experiment (Sections 5, 6, and 7).

   from 1. Introduction

2. The pedagogical review [1] establishes the computational side of the ledger: the Fibonacci anyon model is the simplest anyon model that allows universal quantum computation by braiding alone, and the review works through the full pipeline — braiding matrices, qubit encodings, single-qubit braids approximating target operations to accuracy $\epsilon$ with minimum braid length scaling as $\mathcal{O}(\log(1/\epsilon))$, and the AJL algorithm for evaluating the Jones polynomial — including numerical simulations of the braiding operations.

   from 2. Background and Related Work

3. The paper's method is claim-level verification against primary sources, combined with explicit derivation of every quantitative statement the corrected text retains and an explicitly assumed scaling model for error-correction overhead.

   from 3. Methods

4. This section derives every quantitative statement used by the corrected text.

   from 4. Analysis

What would falsify it

1. A revised readiness assessment of topological quantum computing that follows from the correction, with falsification conditions, limitations, and a smallest de-risking experiment (Sections 5, 6, and 7).

   from 1. Introduction

2. The main corrected claim — that non-Abelian exchange statistics in fractional quantum Hall systems are undemonstrated — would be falsified by a single published interference or fusion-channel measurement that satisfies the operational criterion of Section 4.1: exchange acting as a non-commuting transformation on a degenerate manifold of locally indistinguishable states, per the distinction established in [2].

   from 6. Discussion

3. The overhead result would be falsified if a demonstrated error-correction architecture achieved $10^{2}$–$10^{3}$ logical qubits with materially fewer than $10^{5}$ physical qubits at $p \approx 10^{-3}$, or if the threshold relevant to the full architecture (including the fermion sector) fell below the operating error rate.

   from 6. Discussion

Where the claim stops

1. The quantitative results rest on the following assumptions, with their sensitivity: The braid-length estimate (Section 4.2) assumes a unit proportionality constant in $L = \ln(1/\epsilon)$.

   from 6. Discussion

2. The measurement count (Section 4.3) assumes a unit constant in the $1/\sqrt{N}$ law and a fixed strand number. [1] states that uncertainty grows with the factor $d^{n-1}$ or $d^{n/2-1}$, so the count of $100$ underestimates the iterations needed for larger knots.

   from 6. Discussion

3. The overhead model (Section 4.5) assumes $\alpha = 10^{2}$, $\beta = 1$, $p = 10^{-3}$, $p_{\mathrm{th}} = 10^{-2}$, and is valid only near the stated operating point: as shown in Section 4.5, the linear model does not extrapolate to substantially smaller $p$, where real surface-code behavior departs from it.

   from 6. Discussion

Where to read

1. Abstract229 words · 1 min
2. 1. Introduction522 words · 2 min
3. 2. Background and Related Work1,254 words · 5 min
4. 3. Methods352 words · 2 min
5. 4. Analysis1,099 words · 5 min
6. 5. Results332 words · 1 min
7. 6. Discussion576 words · 3 min
8. 7. Practical Implications and Engineering Pathways634 words · 3 min
9. 8. Conclusion231 words · 1 min
10. Version History88 words · 1 min
11. Provenance and AI Involvement225 words · 1 min