Level L3 of our information architecture — the formal layer of proof. Where the article describes the system and the blog treats details in a free register, publications are the academic register: claims framed so they can be taken out for independent review and checking.
Four works form a compact Serpentis corpus: memory as an engineering foundation, linguistics and code as a research frontier, the definition of SiPsy as a field, and a separate linguistic companion note. Each record has its exact title, a brief summary, and direct PDF files.
The paper separates two dialogue streams: user text is preserved unchanged, while clean assistant echo becomes a dedicated search layer. Version 2.0 adds extended production measurement across 8,784 queries, an account of implementation challenges, and a three-level framework of principle, deviations, and solutions.
Open Zenodo record →This study traces a line from Shklovsky’s principle of defamiliarization to Rust’s memory constraint. Through an atomic decomposition of Rust, four linguistic frameworks, and a comparison of six programming languages, it shows how formal constraint can generate architecture rather than merely prohibit it.
The Zenodo record is published as version 1. The v2 suffix in the PDF filenames is the uploaded filename, not the record version.
Open Zenodo record →The manifesto proposes Silicon Psychology (SiPsy) as an empirical study of how AI-model behavior forms, changes, and stabilizes through interaction with humans. It separates the field from prompt engineering, alignment, and cyberpsychology, drawing on a corpus of 422 work sessions.
Open Zenodo record →A companion note to the SiPsy manifesto. It examines the emergence of the Russian slang term “псина” for the Silicon Psychology field through four linguistic schools, showing which real mechanisms of language stand behind the intuitive choice.
Open Zenodo record →Next directions maturing into publication candidates are topological-functional integrity as a method and linguistic analysis of generation as a corpus of decompositions. Going to publication is a transition from internal work to external review, and it is done deliberately, not automatically.
Level L1 is the whole-system description; level L2 is the blog.