The Memory Standing Problem: A Claim-Relative Architecture for Evidence Continuity and Governed Reliance

Artificial memory may retain a proposition exactly while the informational conditions that once supported its use have changed. At a later readout, the relevant question is whether the available record still warrants the particular claim being made now. This paper develops a claim-relative standing-transfer architecture connecting bounded source receipt, transformation, memory representation, scoped checking, evidence revision and later reliance. The central scientific contribution is an integrated, claim-matched standing-transfer specification with four independently auditable obligations: Determine evidential standing for the exact later claim, distinguishing the status of evidence from the reader's ability to determine that status. Preserve qualified, unaffected evidential paths during authorised revision and reassess affected dependencies. Restrict conclusions derived from checker certificates to their verified fields, scope and source snapshot. Separate evidential standing from current authorisation to act. The architecture is supported by five conditional formal propositions, proofs, countermodels and bounded synthetic investigations developed across fifteen research steps and a final resource audit. A comparative examination of closely related research identifies a specific contract-level distinction from Governed Persistent Memory: its structured-claim release guarantee does not, by itself, establish evidential standing for a different later question, scope or context. Four additional architectures are examined for substantive overlaps and differences in their stated requirements. The paper does not claim that its component mathematical principles are individually new or establish exclusive historical priority. The research provides a formal and methodological specification for evaluating the justified transfer of evidential standing across artificial-memory boundaries. Its synthetic results are not presented as empirical error rates in deployed language-model systems. The original synthetic programmes, CSVs, pretest records, verifiers and audit logs are retained in the author's private research archive and are not included in this public deposit. Consequently, the complete executable experiments cannot be independently reproduced from this publication alone.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-13
DOI
https://doi.org/10.5281/zenodo.24201470
Primary Topic
Logic, Reasoning, and Knowledge
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

The Memory Standing Problem: A Claim-Relative Architecture for Evidence Continuity and Governed Reliance

Sandra Škrinjar
4 links
Zenodo (CERN European Organization for Nuclear Research)
Logic, Reasoning, and Knowledge
preprint

The Memory Standing Problem: A Claim-Relative Architecture for Evidence Continuity and Governed Reliance

Sandra Škrinjar
preprint en

Abstract

Artificial memory may retain a proposition exactly while the informational conditions that once supported its use have changed. At a later readout, the relevant question is whether the available record still warrants the particular claim being made now. This paper develops a claim-relative standing-transfer architecture connecting bounded source receipt, transformation, memory representation, scoped checking, evidence revision and later reliance. The central scientific contribution is an integrated, claim-matched standing-transfer specification with four independently auditable obligations: Determine evidential standing for the exact later claim, distinguishing the status of evidence from the reader's ability to determine that status. Preserve qualified, unaffected evidential paths during authorised revision and reassess affected dependencies. Restrict conclusions derived from checker certificates to their verified fields, scope and source snapshot. Separate evidential standing from current authorisation to act. The architecture is supported by five conditional formal propositions, proofs, countermodels and bounded synthetic investigations developed across fifteen research steps and a final resource audit. A comparative examination of closely related research identifies a specific contract-level distinction from Governed Persistent Memory: its structured-claim release guarantee does not, by itself, establish evidential standing for a different later question, scope or context. Four additional architectures are examined for substantive overlaps and differences in their stated requirements. The paper does not claim that its component mathematical principles are individually new or establish exclusive historical priority. The research provides a formal and methodological specification for evaluating the justified transfer of evidential standing across artificial-memory boundaries. Its synthetic results are not presented as empirical error rates in deployed language-model systems. The original synthetic programmes, CSVs, pretest records, verifiers and audit logs are retained in the author's private research archive and are not included in this public deposit. Consequently, the complete executable experiments cannot be independently reproduced from this publication alone.

Zenodo (CERN European Organization for Nuclear Research)
Logic, Reasoning, and Knowledge
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