2026-10-07T04:06:10Z https://api.figshare.com/v2/oai

oai:figshare.com:article/34016907 2026-09-28T18:16:13Z category_29398 category_29401 category_29416 category_29101 item_type_12 month_year_09_2026

Addendum to «Critical Analysis and Correction Plan for the ARC-22 Model»: Reproducibility Test Results Reen Kuul (25114833) Computational linguistics Corpus linguistics Linguistic structures (incl. phonology, morphology and syntax) Information systems philosophy, research methods and theory ARC-22 Voynich Manuscript reproducibility parser EVA CR <p dir="ltr">This addendum reports an independent reproducibility test of the ARC-22 model, based on the specification published in the critical paper (Kuul, 2026, DOI: 10.6084/m9.figshare.34013346). A Python parser was implemented according to the published rules and run on the ZL3b Voynich corpus (33,206 tokens) and the Timm–Schinner generated text (10,809 tokens). Three parser iterations were performed. None reproduced the target values of CR = 44.38% (Voynich) and CR = 50.50% (generated): observed CR ranged from 37.79% to 63.96% (Voynich) and from 37.63% to 59.98% (generated). The specification is incomplete: slot priority rules, short-token handling, the 32nd kernel, and the formulas for H₂, ΔBPC and LZ77 are not given. Conclusion: ARC-22, as published, does not reproduce its own reported results and requires substantial revision and rebuilding.</p> 2026-09-28T18:16:13Z Text Preprint 10.6084/m9.figshare.34016907.v1 https://figshare.com/articles/preprint/Addendum_to_Critical_Analysis_and_Correction_Plan_for_the_ARC-22_Model_Reproducibility_Test_Results/34016907 CC BY 4.0