The Hidden Advice Problem: Can Adaptive SAT Learning Become a Uniform P = NP Algorithm?

Learning-assisted satisfiability solvers increasingly use pretrained models, dynamically selected heuristics, historical caches, and instance-sensitive preprocessing. Their success introduces a methodological ambiguity: which computational resources belong to the algorithm whose complexity is being claimed? This paper investigates the Hidden Advice Problem, defined as the risk of treating auxiliary information as computationally free while interpreting the resulting online performance as evidence of a uniform worst-case bound. The contribution is a bounded audit protocol, not a new complexity class, theorem, SAT solver, or resolution of P versus NP. It distinguishes four sources of auxiliary information: a single fixed effective program, advice varying with input length, preprocessing specific to the current instance, and state accumulated across earlier episodes. The proposed Cold-Start Uniformity Criterion is deliberately reinterpreted as an evidence and provenance test rather than a demand that every legitimate algorithm repeat its historical development or training. A finite pretrained artifact may be part of an ordinary uniform algorithm; what matters is whether its runtime, representation, precision, and any varying external information satisfy the claimed computational model. Three conceptual figures develop the information-source taxonomy, a global cost-accounting pathway, and a falsification-first evaluation ladder. The study compares this specialization with P/poly, compilability, knowledge compilation, proof complexity, SAT backdoors, contemporary solver engineering, and related construction-obligation research. It concludes that an audit can reject unsupported promotions from empirical speedups to universal guarantees, but no independent advantage over competent existing analysis is demonstrated. No experiments were conducted, no positive polynomial-time algorithm for arbitrary 3-SAT is supplied, and no claim about the truth of P = NP is established.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-19
DOI
https://doi.org/10.5281/zenodo.24371661
Primary Topic
Complexity and Algorithms in Graphs
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

The Hidden Advice Problem: Can Adaptive SAT Learning Become a Uniform P = NP Algorithm?

Rogério Figurelli
Zenodo (CERN European Organization for Nuclear Research)
Complexity and Algorithms in Graphs
preprint

The Hidden Advice Problem: Can Adaptive SAT Learning Become a Uniform P = NP Algorithm?

Rogério Figurelli
preprint en

Abstract

Learning-assisted satisfiability solvers increasingly use pretrained models, dynamically selected heuristics, historical caches, and instance-sensitive preprocessing. Their success introduces a methodological ambiguity: which computational resources belong to the algorithm whose complexity is being claimed? This paper investigates the Hidden Advice Problem, defined as the risk of treating auxiliary information as computationally free while interpreting the resulting online performance as evidence of a uniform worst-case bound. The contribution is a bounded audit protocol, not a new complexity class, theorem, SAT solver, or resolution of P versus NP. It distinguishes four sources of auxiliary information: a single fixed effective program, advice varying with input length, preprocessing specific to the current instance, and state accumulated across earlier episodes. The proposed Cold-Start Uniformity Criterion is deliberately reinterpreted as an evidence and provenance test rather than a demand that every legitimate algorithm repeat its historical development or training. A finite pretrained artifact may be part of an ordinary uniform algorithm; what matters is whether its runtime, representation, precision, and any varying external information satisfy the claimed computational model. Three conceptual figures develop the information-source taxonomy, a global cost-accounting pathway, and a falsification-first evaluation ladder. The study compares this specialization with P/poly, compilability, knowledge compilation, proof complexity, SAT backdoors, contemporary solver engineering, and related construction-obligation research. It concludes that an audit can reject unsupported promotions from empirical speedups to universal guarantees, but no independent advantage over competent existing analysis is demonstrated. No experiments were conducted, no positive polynomial-time algorithm for arbitrary 3-SAT is supplied, and no claim about the truth of P = NP is established.

Zenodo (CERN European Organization for Nuclear Research)
Complexity and Algorithms in Graphs
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon