Research notes
This page holds exploratory material: drafts, circulating working papers, and lines of argument still in preparation. None of it is peer-reviewed, and none of it belongs to the publication record. Items move to Publications only once a journal decision has been made.
Theme 01
Principal current manuscript. Identifies the institutional site at which answerability must be borne once reasoning is delegated to automated procedures, and shows how nominal oversight can leave that site vacant.
Introduces second-order provenance: the traceability not of data but of the normative orderings under which data are taken to matter.
Specifies four conditions on a maintenance license — the entitlement to treat a predictive output as a ground for a given intervention — and shows that predictive accuracy does not by itself confer any of them.
Case · AI-driven predictive maintenance in industrial asset management serves as the applied test case.
Read the abstract & argument →
Argues that answerability requires an addressable record — not merely an explainable output — and reconstructs institutional recordkeeping as a moral infrastructure.
Theme 02
Studies how ontological commitments organise the space in which normative orderings can be constructed at all.
Argues that predictive systems can enact a second-personal exclusion: those about whom claims are made are structurally not among those to whom answer is owed.
Theme 03
Analyses how mimetic structures of the self are recomposed when models of the self are precomputed for us.
Theme 04
Describes the platforming of desire: the conversion of desire into a resource whose comparative structure can be measured, ranked, and monetised.
Theme 05
Provides a map of epistemic instabilities that arise where model, world, and design fail to align.
Formulates a criterion of design–inference alignment: the inferences licensed by an empirical design must not exceed the constitution of its object.
Sketches a general account of how the constitution of scientific objects constrains the inferences that can be drawn about them.
Diagnoses ontological assumptions smuggled in by Markov-model formulations of state and transition.
Shows why the standard experimental repertoire around large language models systematically overreaches its inferential license.
Theme 06
Argues that AI systems function as epistemic infrastructure — silent scaffolding of what an institution can know — and reframes safety and governance in those terms.
Describes institutional inversion: the pattern by which supporting systems come to determine the institutional practices they were built to serve.
Reconstructs friction as a productive, not merely obstructive, feature of institutional systems and shows what its removal costs.
Proposes an organic account of institutional interfaces against a purely mechanical API metaphor.
Argues that accumulated administrative expertise can itself become a source of epistemic lag under rapid technological change.
Treats media systems as infrastructure whose formative effects operate below the level of content.