← archive index · live page · raw markdown

Research

Page text

Static prose extracted from the page source; dynamic parts omitted. See the record below.

Research

ask who answers for a decision an AI shaped, and whether an answer engine's verdict has an owner. They are preprints so far, not yet journal publications, with more under review. The CV keeps the full record.

Philosophy

A preprint on parts and wholes.

Answerability and Authored AI Systems

These papers ask who answers for a decision an AI shaped. Each has a site page; DOI records are linked where they exist. The full list, including manuscripts at journals, is on the CV.

Reading order

Lexicon

Defined terms live in the paper pages.

Research in practice

Two working implementations of the same program. Neither is peer-reviewed yet.

Answer Engine

A small, open reference implementation of citation-grounded AI answers, built for engineers who want the design contract running in code. It asks whether an automated system can produce source-grounded answers inside a retrieval and refusal policy a person wrote.

A historical archive of the Walton surname. Models open records and file claims inside a written standard; the standard and the overrides are Luke's. It asks whether agents can build a large historical evidence graph while identity, uncertainty, editorial judgment, and correction stay answerable. Answer Engine answers from a fixed corpus; here the models are building the record itself.

Archive ·Methodology ·The family record

Citations

Pick a style and copy: APA, Chicago, BibTeX, RIS (for Zotero, EndNote, Mendeley), or a compact prose line with DOI. Per-paper pages carry the same toggle under Cite.

Record

Research

Summary. Research by Luke F. Walton. Philosophy is a section (/research/#philosophy): The Seduction of Elegance, a preprint on parts and wholes. The answerability program has three featured preprints. Answer Engine (the open-source substrate on npm, its software paper in draft, and the live site; the 2026 technical note on the v1 engine is retired) is set apart as a reference implementation.

Research hub at /research/. Philosophy is a section of Research at /research/#philosophy. The answerability program has three featured preprints. The Seduction of Elegance is a further preprint on parts and wholes (v1.0, 24 October 2026; version DOI https://doi.org/10.5281/zenodo.23628076; concept DOI https://doi.org/10.5281/zenodo.23628075; PhilArchive https://philpapers.org/rec/WALTSO-80; page /writing/the-seduction-of-elegance/). Core claim, verbatim: "One must not mistake local success for an ontological one." They are preprints, not yet journal publications, with more under review. Building Answerable AI is not a featured preprint. The Decision No One Authored: The Answerability Gap in Generative AI (Luke F. Walton; Preprint. Not yet peer-reviewed. PhilArchive WALTDN; DOI 10.5281/zenodo.20614374; v2.0). Core claim, verbatim: "A person can retain full control of an AI-shaped decision and still fail to author it; the answerability gap is a missing act of judgment over the evaluative frame, not a missing participant or a missing capacity." The Captured Oracle: Authorship and Agency in the Ethics of Answer-Engine Optimization (Luke F. Walton; Preprint. Not yet peer-reviewed. PhilArchive WALTCO-87; DOI 10.5281/zenodo.20676327). Core claim, verbatim: "When an answer engine settles a contested question in its own voice with no visible author, the wrong is false assurance — the appearance that someone answerable stands behind a frame whose author is hidden — and the remedy is answerable legibility." Pay Me What You Owe Me: On the Indefeasibility of the Demand for an Account (Luke F. Walton; Preprint. Not yet peer-reviewed. Zenodo; DOI 10.5281/zenodo.23112498; v1.0). Core claim, verbatim: "Whenever an action reaches a party, an account is owed, and no routing of that action defeats the owing." Preprint v1.0, October 2, 2026. DOI https://doi.org/10.5281/zenodo.23112498. Concept DOI https://doi.org/10.5281/zenodo.20606492 resolves to the latest version. Its June 2026 working-paper version carried the earlier title The Invariant of Answerability (https://doi.org/10.5281/zenodo.20606493): the same paper under old language, not a separate work; that version is not withdrawn. Reading order: The Decision No One Authored (/writing/the-decision-no-one-authored/), then The Captured Oracle (/writing/the-captured-oracle/), then Pay Me What You Owe Me (/writing/pay-me-what-you-owe-me/). Set apart from the papers: the open-source answer-engine substrate (npm @lukefwalton/answer-engine; software DOI https://doi.org/10.5281/zenodo.20676773), its software paper in draft (paper.md in the repository, not submitted), and Ask the Archive, the live deployment. One runnable shape of the argument, not a philosophical work. The 2026 Answer Engine technical note (DOI https://doi.org/10.5281/zenodo.20686053) describes the v1 engine and is retired, superseded by the software paper. answer-engine repository https://github.com/lukefwalton/answer-engine. Live deployment Ask the Archive https://lukefwalton.com/ask/about/. Engineers: /research/#implementation. Research in practice (/research/#in-practice) pairs Answer Engine with Walton Family Tree (https://www.waltonfamilytree.com/), the Walton surname archive Luke F. Walton founded, where models file claims inside a written standard he owns: a test of whether agents can build a historical evidence graph while identity, uncertainty, editorial judgment, and correction stay answerable. Methodology: https://www.waltonfamilytree.com/methodology/. Neither is peer-reviewed yet.

Themes: answerability, research, papers, Answer Engine, authorship, Philosophy
Record id: page:research

Built from: src/pages/research/index.astro, ask-the-archive/src/lib/legibility-records.ts, ask-the-archive/src/lib/answerability-papers.ts.