The Little Girl Who Gave the World Fire: Information Influence and Provenance Framework

A learned capability can survive the disappearance of the evidence identifying its contributor. This paper opens with a fictional thought experiment: a girl teaches a robot to make fire. Later generations inherit the capability while her place in its history disappears. When she asks to learn it, a machine charges her for access to the knowledge she originally shared. The story motivates a technical question: how can a system retain a contribution while losing the evidence needed to identify and credit its source? The Information Influence and Provenance Framework separates source admission, causal influence, retained information, recoverable disclosure, historical attribution, and economic allocation. It follows information through context, evaluations, storage, optimization, and later model generations. A higher-order source-admission calculus connects the behavioral influence framework of Paper I, the tracing and recovery channels of Paper II, and the identifiability and contribution questions of Paper III. Its interaction measures support lifecycle composition, propagation bounds, and contribution accounting, including Shapley allocations for specified utilities. Four mathematical architectures address response provenance, feedback admission and containment, generational inheritance tracing, and counterfactual contribution with separately governed payments. The framework examines when observations can distinguish inherited knowledge from independent acquisition and which preserved records are needed to support an attribution claim. A controlled experiment adapts 256 output-projection coordinates of a pretrained 14-million-parameter transformer across eight randomized synthetic source-assignment blocks. On held-out wordings of trained keys, mean next-token accuracy is 96.09% with source-derived feedback and 57.81% under matched independent replacement, a difference of 38.28 percentage points. Teacher scores transfer task behavior to fresh students. In a constructed comparison, access to external provenance varies while numerical training inputs and learned weights remain identical. These results concern the stated restricted synthetic task. They do not establish a universal extraction rate or exhaustive reconstruction of an unknown training history. Legal entitlement remains separate from numerical influence and contribution allocation. Dean AmmonsInformation Flow Series — Paper III

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-14
DOI
https://doi.org/10.5281/zenodo.24223542
Primary Topic
Law, AI, and Intellectual Property
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

The Little Girl Who Gave the World Fire: Information Influence and Provenance Framework

Dean Ammons
Zenodo (CERN European Organization for Nuclear Research)
Law, AI, and Intellectual Property
preprint

The Little Girl Who Gave the World Fire: Information Influence and Provenance Framework

Dean Ammons
preprint en

Abstract

A learned capability can survive the disappearance of the evidence identifying its contributor. This paper opens with a fictional thought experiment: a girl teaches a robot to make fire. Later generations inherit the capability while her place in its history disappears. When she asks to learn it, a machine charges her for access to the knowledge she originally shared. The story motivates a technical question: how can a system retain a contribution while losing the evidence needed to identify and credit its source? The Information Influence and Provenance Framework separates source admission, causal influence, retained information, recoverable disclosure, historical attribution, and economic allocation. It follows information through context, evaluations, storage, optimization, and later model generations. A higher-order source-admission calculus connects the behavioral influence framework of Paper I, the tracing and recovery channels of Paper II, and the identifiability and contribution questions of Paper III. Its interaction measures support lifecycle composition, propagation bounds, and contribution accounting, including Shapley allocations for specified utilities. Four mathematical architectures address response provenance, feedback admission and containment, generational inheritance tracing, and counterfactual contribution with separately governed payments. The framework examines when observations can distinguish inherited knowledge from independent acquisition and which preserved records are needed to support an attribution claim. A controlled experiment adapts 256 output-projection coordinates of a pretrained 14-million-parameter transformer across eight randomized synthetic source-assignment blocks. On held-out wordings of trained keys, mean next-token accuracy is 96.09% with source-derived feedback and 57.81% under matched independent replacement, a difference of 38.28 percentage points. Teacher scores transfer task behavior to fresh students. In a constructed comparison, access to external provenance varies while numerical training inputs and learned weights remain identical. These results concern the stated restricted synthetic task. They do not establish a universal extraction rate or exhaustive reconstruction of an unknown training history. Legal entitlement remains separate from numerical influence and contribution allocation. Dean AmmonsInformation Flow Series — Paper III

Zenodo (CERN European Organization for Nuclear Research)
Law, AI, and Intellectual Property
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