REVIEW 3 major objections 6 minor 2 references
The Aurellion Function: A Recursive Fast-Growing Hierarchy Beyond Knuth Notation
T0 review · 3 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash
Pith's one-line read The paper defines the Aurellion function $A_1=10\uparrow\uparrow\uparrow 10$, $A_{n+1}=10\uparrow^{A_n}10$, and claims it dominates every function provably total in Peano Arithmetic.
desk verdict A Graham-number-style sequence with base 10; the main domination claim is false and unsupported. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The machinery is a self-referential hyperoperation: starting from $A_1 = 10 \uparrow\uparrow\uparrow 10$, each next value puts the previous value into the arrow index, $A_{n+1} = 10 \uparrow^{A_n} 10$. This makes the 'height' of the operation grow with the function's own output rather than with the input $n$, which the paper uses to try to diagonalize over all finite up-arrow levels in a single sequence. The comparison to the fast-growing hierarchy $f_\alpha$, defined by iterating functions at successor stages and taking fundamental sequences at limit stages, is the yardstick by which the paper measures the claimed domination.
What would settle it
A direct comparison with the fast-growing hierarchy would settle the point: if one can show $A_{n+1} \le f_\omega(A_n)$ for all sufficiently large $n$, then $A_n \le f_\omega^n(A_1)$, which is in turn bounded by $f_{\omega+1}(n)$ for large $n$. Since $f_{\omega+1}$ is itself provably total in Peano Arithmetic and lies at an ordinal stage below $\varepsilon_0$, such a bound would contradict the claim that $A_n$ dominates every $f_\alpha$ with $\alpha<\varepsilon_0$.
Extended reading notes
Core claim
The central claim is that $A_n$ is a total computable function that outgrows $f_\alpha$ for every $\alpha < \varepsilon_0$ in the fast-growing hierarchy. Since the functions provably total in Peano Arithmetic are exactly those bounded by such $f_\alpha$ below $\varepsilon_0$, the paper concludes that $A_n$ dominates all PA-provably total functions and that PA cannot prove the totality of $A_n$. It goes further and situates the function near the proof-theoretic ordinal $\Gamma_0$, suggesting that totality is provable in systems such as $\mathrm{ACA}_0$ or $\mathrm{ID}_1$ rather than in PA. The argument for the domination claim is heuristic: it reasons that because the arrow count in $A_{n+1} = 10 \uparrow^{A_n} 10$ is itself enormous, the sequence must eventually outrun any fixed ordinal stage $\alpha < \varepsilon_0$.
Load-bearing premise
The load-bearing assumption is that using the previous enormous value as the next arrow count is equivalent to diagonalizing over all ordinals below $\varepsilon_0$; the paper presents this as a heuristic and leaves the required ordinal-notation embedding for future work.
Editorial extensions
If this is right
- If the domination claim holds, $A_n$ is a single computable function that marks the boundary of what Peano Arithmetic can prove total.
- The sequence would give a concrete, easily defined witness that PA is incomplete with respect to totality statements.
- The ordinal-indexed extension $A_\alpha$ would provide a hierarchy of large numbers that connects recursive constructions with proof-theoretic ordinals.
- Because $A_n$ is defined without explicit ordinal notations, it would be an example of a function at the claimed strength level arising directly from a hyperoperation recursion.
Reading between the lines
- Under a standard ordinal analysis, the recurrence $A_{n+1}=10\uparrow^{A_n}10$ is only one application of the limit level $f_\omega$ to the previous value, so the sequence would naturally land near $f_{\omega+1}$ rather than $\varepsilon_0$; the paper's heuristic turns rapid numerical growth of the index into an assumed transfinite diagonalization.
- The proposed limit definition $A_\lambda = \sup_n A_{\lambda[n]}$ depends on a choice of fundamental sequences for each limit ordinal, so it is not a canonical construction until a notation system is fixed.
- A weaker, testable version of the conjecture is whether $A_n$ eventually dominates every fixed finite level $f_k$; that statement can be checked with explicit bounds even if the full claim about all $\alpha<\varepsilon_0$ fails.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper defines a function A by A_1 = 10 ↑↑↑ 10 and A_{n+1} = 10 ↑^{A_n} 10, proves a lower bound A_n ≥ 10 ↑^{n+2} 10 in Lemma 4.1, and claims that A_n dominates every f_α for α < ε0, hence all functions provably total in Peano Arithmetic, and that the function is therefore located near the proof-theoretic ordinal Γ0. The domination claim is explicitly labeled a conjecture in Section 4.2, where a formal proof is deferred to future work, but the abstract and Section 5.2 state the same claim as an established result.
Significance. The paper is transparent about its definition and proves one correct elementary lower bound. If the main claim were true, the Aurellion function would be a surprisingly simple recursive function calibrating the strength of PA and beyond, which would make the paper a useful bridge between large-number hierarchies and ordinal analysis. The central claim is nonetheless false: standard fast-growing-hierarchy estimates place A_n at the level of f_{ω+1}, far below f_{ω^2}, which is already PA-provably total. The paper cannot fulfill its main advertised contribution without a substantive change to the definition or claims; the definition and the lower-bound lemma do not by themselves constitute the promised proof-theoretic calibration.
major comments (3)
- [Abstract and §5.2] The claim that A_n dominates all functions provably total in Peano Arithmetic is false. For each fixed m, 10 ↑^m 10 is bounded by f_ω(m+C) for a suitable constant C, because f_ω(k)=f_k(k) majorizes any fixed hyperoperation. The recurrence then gives A_{n+1}=10 ↑^{A_n}10 ≤ f_ω(A_n+C), and iterating the bound yields A_n ≤ f_ω^n(A_1+C) ≤ f_{ω+1}(n+D) for all sufficiently large n, with D a fixed constant. Since f_{ω^2} is provably total in PA (ω^2 < ε0) and eventually dominates f_{ω+1}, A_n cannot dominate all PA-provably total functions. Moreover, f_{ω+1} itself is PA-provably total, so the companion assertion in §5.2 that PA cannot prove the totality of A_n is also incorrect, assuming the bound above is formalizable.
- [§4.1–4.2] The heuristic argument for Conjecture 4.2 is not a proof and does not support the domination claim. Lemma 4.1 only shows that A_n outgrows each fixed finite hyperoperation level; it says nothing about all ordinal levels below ε0, since those levels are not exhausted by finite arrow counts. The deferred 'precise embedding into an ordinal notation system' is exactly the missing step, and the natural attempt at such an embedding is blocked by the bound in the previous comment, which places A_n at the level of f_{ω+1} rather than cofinal in ε0.
- [Abstract and §8] The conclusion that the Aurellion function is 'near Γ0' does not follow even from the false domination claim. If A_n did majorize every f_α for α < ε0, that would calibrate it at the ε0 level, not at Γ0, which is a much larger ordinal associated with predicative systems. The abstract and conclusion present the PA-domination and Γ0 placement as established facts, while §4.2 explicitly labels such a comparison a conjecture requiring future work; these internal inconsistencies need to be resolved.
minor comments (6)
- [§4.1] The statement that f_ω(n) corresponds roughly to iterated exponential growth is inaccurate: f_ω diagonalizes the finite levels of the fast-growing hierarchy and is Ackermann-like, while iterated exponential growth is the level of f_3.
- [Appendix A] The editorial note 'This section is redundant ... I would remove this appendix section' must not appear in a published manuscript; either remove the appendix or remove the note.
- [§2.2] The definition of fundamental sequences is incomplete; for example, the clause for λ=ω^α with α a limit ordinal should specify λ[n]=ω^{α[n]}, and a general treatment for ordinals below ε0 is needed for the hierarchy to be fully well-defined.
- [§5.1] The remark that the numeric value of A_n 'cannot be explicitly computed or stored for even small n' is informal and adds little; if kept, it should be phrased in terms of the size of the decimal representation rather than as a statement about computability.
- [§5.2] The description of ACA0 as 'Arithmetical Comprehension Axiom with ω-iteration' is nonstandard and misleading; ACA0 has proof-theoretic ordinal ε0 and does not literally contain an ω-iteration axiom.
- [References] The text credits Löb as well as Wainer for the fast-growing hierarchy, but only Wainer's paper is listed; a citation to the Löb–Wainer paper should be added if the attribution is retained.
Circularity Check
No circularity: the paper's unsupported Section 5.2 dominance claim is an omitted proof, not a reduction of a conclusion to its own inputs.
full rationale
The Aurellion paper contains no step in which a conclusion is made identical to its premise by definition, no fitted parameter is relabeled as a prediction, and no load-bearing self-citation is used. The definition A_{n+1}=10 ↑^{A_n}10 is an ordinary recursive construction, and Lemma 4.1 derives a lower bound from it without circularity. The central assertion that A_n dominates every f_α for α<ε_0, and hence all PA-provably total functions, is presented in Section 5.2, but the paper's own Section 4.1 labels the corresponding statement as Conjecture 4.2 and explicitly states that a formal proof requires an ordinal-notation embedding 'which is part of future work.' Promoting a conjecture to an established result is a missing-proof defect, not a circular derivation: the conjecture is not built into the definition of A_n by construction, and the heuristic comparison in Section 4.1 does not force the dominance claim through any equation that reduces to itself. A standard fast-growing-hierarchy estimate would in fact refute the claim, but that is a correctness problem, not circularity. Accordingly, no circular step is identified and the score is 0.
Assumptions & free parameters
assumptions (4)
- standard math Knuth up-arrow is total and strictly increasing in arrow count for bases above 1
- standard math The fast-growing hierarchy f_alpha is well-defined for all alpha below epsilon_0 via fundamental sequences
- domain assumption Functions provably total in PA are exactly those majorized by some f_alpha with alpha below epsilon_0
- ad hoc to paper The heuristic that A_n's self-referential arrow count lets it outgrow every fixed ordinal below epsilon_0
Cite this review
Pith. "Pith review of The Aurellion Function: A Recursive Fast-Growing Hierarchy Beyond Knuth Notation." pith.science (2026). https://pith.science/paper/YDQ37G4Q
@misc{pith2026250605067,
author = {Pith},
title = {Pith review of: The Aurellion Function: A Recursive Fast-Growing Hierarchy Beyond Knuth Notation},
year = {2026},
howpublished = {\url{https://pith.science/paper/YDQ37G4Q}},
note = {Machine review of arXiv:2506.05067}
}
abstract
We introduce the Aurellion Function, a novel recursively defined fast-growing hierarchy based on Knuth's up-arrow notation, defined by $A_1 = 10 \uparrow\uparrow\uparrow 10$, $A_{n+1} = 10 \uparrow^{A_n} 10$, where the number of arrows in the operation increases superexponentially with $n$. We analyze its growth rate relative to classical hierarchies such as the fast-growing hierarchy $(f_\alpha)_{\alpha < \varepsilon_0}$, and discuss its provability status in formal arithmetic. We provide formal bounds showing $A_n$ dominates all functions provably total in Peano Arithmetic, situating the Aurellion Function near the proof-theoretic ordinal $\Gamma_0$ due to its ability to majorize all functions $f_\alpha$ for $\alpha < \varepsilon_0$. We also outline possible transfinite extensions indexed by countable ordinals, thus bridging symbolic large-number constructions and ordinal analysis.
Reference graph
Works this paper leans on
Reviewed August 7, 2026 · model on record in the stance chip above.
Discussion (0). Sign in to comment.