Underground Information Architecture Against Systematic Non-Retrieval: NPH, the Censorship Observatory, and the De-Designer Function
This essay presents, in critical edition, the verbatim dialogue between the author ("gk"),¹ architect of a 25-million-word voice-transcription corpus, and Lumo,² the Proton AI assistant serving as auditor and design collaborator, across sessions conducted 29–30 October 2026 on the machine designated "mn w."³ The dialogue documents the founding of a "super-search" project — an attempt to design information-retrieval infrastructure that routes around demonstrated retrieval failures — and the simultaneous construction of a "censorship observatory" that instruments instances of non-retrieval. The edition preserves the participants' exact wording. Spoken turns (transcribed by machine from voice input) have been normalized for punctuation and typographical error only, with every substantive correction recorded in the footnotes.⁴ Analytical commentary accompanies the transcript, examining the epistemic discipline the dialogue itself established: the attribution doctrine (motive documented, act measured, attribution held prima facie against a pool rather than a party), the collapse of the economic-ranking/censorship distinction, and the emergence of the "de-designer" as an unowned project function. The edition argues that the dialogue's chief methodological achievement is not any single finding but the sustained separation of what was measured from what was inferred — a discipline maintained asymmetrically against both the platforms under study and the participants' own case file.
Authors
- Kasey George (ORCID: https://orcid.org/0009-0005-5487-7435)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-10-31
- DOI
- https://doi.org/10.5281/zenodo.23840535
- Primary Topic
- Ethics and Social Impacts of AI
- Type
- preprint