Artificial Systems as an Engineered Reference Class for Recursive Dimensional Creation: Formal Limits on Origin Identification and Objective Preservation During Model Revision

How much information about an upstream system, including its identity and originating objective, can be recovered by observers restricted to a downstream interface? Human-engineered artificial systems provide an accessible reference class: designers specify substrates, objectives, and constraints; downstream systems act within those conditions; and selected results return through observation channels. This article formalizes a recursive upstream-downstream architecture and applies established point-, set-, systems-, and interventional-identification concepts to origin attribution. A non-injective observation map partitions upstream states into equivalence classes. An upstream property is point identified only if it is constant on the relevant class; a matched alternative sharing the complete accessible law prevents point identification at the declared interface. Nine linked empirical and formal stages examine inherited state under environmental shift, objective-and-constraint governance during model revision, attribution under corruption and observer selection, local-to-global compatibility, cross-domain correspondence, and registry measurement. The record preserves negative results: inherited information did not guarantee adaptation; provenance markers changed none of 144 tested actions; the observer-selection correction failed; and the registry's real-packet reliability gate blocked confirmatory scoring. The results establish an engineered reference class, component-level feasibility, and an interface-relative identified set. They neither identify an intelligent cosmological origin nor provide comparative likelihood support among registered alternatives. The framework instead specifies what additional observations or interventions would be required to refine the identified set. This v3.11 deposit also includes Reproducibility and Research Record Supplement v2.1, the consolidated append-only research record through RR75, Systematic Novelty and Prior-Art Review v3.10, editorial and integrity audits, the RR1–RR5 and RR44–RR45 reproducibility artifact subset, and SHA256SUMS.txt. A 482-entry inventory identifies the complete Experiment 1 production archive; the 120 full trajectories, 360 checkpoints, remaining underlying RR1–RR52 files, and specified Experiment 9 development directories remain in the controlled archive. This preprint has not undergone journal-administered peer review. Licensing: Non-code materials are licensed under CC BY 4.0. The five Python source files identified in SOFTWARE_LICENSE_MIT.txt are licensed separately under the MIT License. Third-party dependencies retain their own licenses.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-14
DOI
https://doi.org/10.5281/zenodo.24222831
Primary Topic
Complex Systems and Dynamics
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Artificial Systems as an Engineered Reference Class for Recursive Dimensional Creation: Formal Limits on Origin Identification and Objective Preservation During Model Revision

Marvin Shawn Ezzell
Zenodo (CERN European Organization for Nuclear Research)
Complex Systems and Dynamics
preprint

Artificial Systems as an Engineered Reference Class for Recursive Dimensional Creation: Formal Limits on Origin Identification and Objective Preservation During Model Revision

Marvin Shawn Ezzell
preprint en

Abstract

How much information about an upstream system, including its identity and originating objective, can be recovered by observers restricted to a downstream interface? Human-engineered artificial systems provide an accessible reference class: designers specify substrates, objectives, and constraints; downstream systems act within those conditions; and selected results return through observation channels. This article formalizes a recursive upstream-downstream architecture and applies established point-, set-, systems-, and interventional-identification concepts to origin attribution. A non-injective observation map partitions upstream states into equivalence classes. An upstream property is point identified only if it is constant on the relevant class; a matched alternative sharing the complete accessible law prevents point identification at the declared interface. Nine linked empirical and formal stages examine inherited state under environmental shift, objective-and-constraint governance during model revision, attribution under corruption and observer selection, local-to-global compatibility, cross-domain correspondence, and registry measurement. The record preserves negative results: inherited information did not guarantee adaptation; provenance markers changed none of 144 tested actions; the observer-selection correction failed; and the registry's real-packet reliability gate blocked confirmatory scoring. The results establish an engineered reference class, component-level feasibility, and an interface-relative identified set. They neither identify an intelligent cosmological origin nor provide comparative likelihood support among registered alternatives. The framework instead specifies what additional observations or interventions would be required to refine the identified set. This v3.11 deposit also includes Reproducibility and Research Record Supplement v2.1, the consolidated append-only research record through RR75, Systematic Novelty and Prior-Art Review v3.10, editorial and integrity audits, the RR1–RR5 and RR44–RR45 reproducibility artifact subset, and SHA256SUMS.txt. A 482-entry inventory identifies the complete Experiment 1 production archive; the 120 full trajectories, 360 checkpoints, remaining underlying RR1–RR52 files, and specified Experiment 9 development directories remain in the controlled archive. This preprint has not undergone journal-administered peer review. Licensing: Non-code materials are licensed under CC BY 4.0. The five Python source files identified in SOFTWARE_LICENSE_MIT.txt are licensed separately under the MIT License. Third-party dependencies retain their own licenses.

Zenodo (CERN European Organization for Nuclear Research)
Complex Systems and Dynamics
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