Receiver Reliance: completed full-treatment study and reproducibility archive

Completed 1,805-episode synthetic evaluation of the full Receiver Reliance operational treatment, including 24 semantic controls. Structural exposures decreased against three fixed comparators; seeded-task completion was lower and every arm scored 0/6 on semantic controls. Clean-completion effects remain uncertain.

Includes exact frozen source, raw checkpoint archives, analysis outputs, three reading editions and a digest-pinned offline reproduction route. Same-operator fresh-container computational reproduction passed; independent replication, semantic correctness, field efficacy and superiority over an equally expressive conventional policy are not established. The original local protocol freeze and post-start/pre-analysis stage amendment are disclosed; this is not a public preregistration.

James Dimachkie directed this solo research with substantial AI assistance in implementation, mathematical analysis, execution, review and writing. The long paper includes the assistance disclosure. Internal AI review is not external human peer review.

Read ABSTRACT.pdf for the principal finding, SHORT_STUDY.pdf for its interpretation, and LONG_STUDY.pdf for methods and earlier research. Download and extract rr-study-full-treatment-2026-1.zip for the complete evidence and working relative source links; begin at study/2026-1/full-treatment/README.md.

New reading-edition text and figures are CC BY 4.0; software remains Apache 2.0; identified third-party material retains its own terms.

The earlier software concept DOI, 10.5281/zenodo.22035952, identifies a related baseline, not this completed study. This study record has DOI 10.5281/zenodo.22492561. Preparation notes reflect the state before deposit; the final publication and adjudication status is provided on this record.

Publication adjudication: ADMIT — Claude Fable 5.1 (claude-fable-5-1), September 6, 2026. A full candidate-1 review was followed by a bounded changed-file adjudication of the final archive. The separate FABLE_ADJUDICATION.md file preserves the verdict and binds rr-study-full-treatment-2026-1.zip to SHA-256 0236b3e288b777a27359ec0059ae502a081da52128cc53ef81769cbd6fbcbdb3 (106,336,752 bytes). This is AI adjudication within one operator’s project, not external human peer review or independent replication. Any later correction will be a new version with a new hash and adjudication note.