PROXY SEDUCTION FRAMEWORK

Research Hub

Working hub for the empirical phase, organized along the research timeline rather than by artifact type. The sections below follow the programme arc: the framework and first-year assessment work, Study 1 fieldwork in software development, analysis and framework refinement, Study 2 multi-domain fieldwork, and the paper and dissertation pipeline. Each card links to a standalone page. Placeholder cards mark work that will be built as the programme progresses.

Study 1 target: ~50 interviews, 70/30 frontline to boundary Primary domain: professional software development FY Report: due 28 Aug 2026 Arc: Paper 1 (Jun 2026) → dissertation (Jul 2028)
PSF Theory Hub →
01 Framework & First-Year Assessment 6 items
FRAMEWORK
Literature Synthesis
The working synthesis behind the framework (updated 9 Jun 2026). Traces the move from Evaluative Discontinuity Theory to PSF, the four theoretical resources, the three dimensions of evaluative capacity, the evidence constellation, and the cross-cutting causal chain. Authoritative reference for each source's role in the framework.
7 sections EDT → PSF Working, not settled
Active
FRAMEWORK
Literature Review
The FYR literature review built as a problematization (Sandberg and Alvesson 2011). A structured Scopus search across four blocks, then the continuity assumption surfaced across four literatures (sociomateriality, institutional logics, dynamic capabilities, epistemic regimes), and the resources PSF integrates across levels to name what each literature leaves out.
Problematization 4 literatures ~2,900 words
Active · drafting
FY ASSESSMENT
FY Report: Annotated Outline
Draft table of contents and scope statement for the First Year Report, the pre-reading for the FYA1 preliminary meeting. Eleven sections with planning word budgets, totalling approximately 14,000 words including appendices. Working title on organizational evaluative capacity under transformative AI engagement.
~14,000 words Due 28 Aug 2026 FYA1 pre-read
Active · drafting
FY ASSESSMENT
FY Report: Methodology Account
Reflexive account of the four design choices that carry the most weight: the discrimination test reads reachability rather than spontaneous loss, the pre-engagement baseline is reconstructed rather than sampled, sampling sorts on observables with the prediction held separate, and the 70/30 split is a judgment under uncertainty. Each choice stated with its rejected alternative.
4 design choices Reflexive Rejected alternatives
Active · drafting
FY ASSESSMENT
First-Year Logbook
Record of researcher development across the first three terms (Michaelmas 2025, Lent and Easter 2026): 21 Researcher Development Courses plus the ISO5 qualitative methods module at Cambridge Judge. Tracks the training that supports the empirical phase and the FY assessment record.
21 RDCs ISO5 module 3 terms
Active
PLANNING
Research Timeline (Gantt)
Three-year programme map. Lit synthesis and framework run through Year 1, then Paper 1 (Jun 2026) and the FY Report (Aug 2026), Fieldwork Phase 1 into data analysis and framework refinement, Paper 2 (Mar 2027), Fieldwork Phase 2 and synthesis, Paper 3 (Dec 2027), then dissertation writing to submission (Jul 2028).
3 years 3 papers 2 fieldwork phases
Active
02 Study 1 Fieldwork · Software Development 5 items
FIELDWORK · MASTER DOCUMENT
Study 1: Research Design
The governing design document for Phase 1 fieldwork. Sets the two screening axes (consequence exposure and AI engagement depth), the four-cell structure read through two strictly separated frames (screening versus coding), the research questions (RQ1 judgment, RQ2 propagation), the scope against Paper 1's three directions, the sample size and allocation (about 50 interviews, 70/30 frontline to boundary), the two-interview protocol, and the falsification spine. The interview protocol, coding rubric, and empirical checklist cards below all derive from this document.
2 axes, 4 cells RQ1 + RQ2 ~50 interviews Screening vs coding frame ~8,900 words
Active
FIELDWORK · INSTRUMENT
Interview Protocol
Two semi-structured interviews that observe one typification loop at two points: the frontline interview (object: the code shipped) and the boundary-activity interview (object: the metric passed upward). Each runs about fifty minutes as a linear flow, with three scripted moves. Neither names AI, judgment, proxies, criteria, or erosion. Rebuilt from the Study 1 design.
2 interviews 3 moves each Reachability probe
Active
FIELDWORK · INSTRUMENT
Coding Rubric
The transcript-level rubric that operationalizes the falsification spine. Built around the Position 1/2/3 discrimination test (unprompted, self-supplied real case, interviewer-supplied hypothetical), which is what separates proxy seduction from skill atrophy. Confirming and disconfirming signatures for RQ1 and RQ2. Applied in analysis only, never visible to participants.
Position 1/2/3 Confirm + disconfirm RQ1 + RQ2
Active
FIELDWORK · INSTRUMENT
Empirical Phase Checklist
Operational checklist for running Phase 1 end to end: screening and cell allocation against targets, ethics and consent state, the per-interview run sheet, the no-collection and no-retention handling of anchored work examples, post-interview memo and coding steps, and the saturation and frame-separation disciplines. Rebuilt from the Study 1 design and methodology.
Screening → coding Ethics state Run sheet
Active
FIELDWORK · PARTNERSHIP
Joint Fieldwork: Microsoft
Planned joint interview study with Steven Clarke (Principal UX Researcher, Microsoft) on how AI tools reshape developers' sense of success at work. Microsoft recruiting identifies and schedules participants, interviews and analysis are joint, and the findings inform both Microsoft product teams and the PSF academic work. Pending legal and ethics approvals from Microsoft and Cambridge.
With Steven Clarke Approvals pending IP: publishable
Pending approvals
03 Analysis & Framework Refinement 2 planned
ANALYSIS
Phase 1 Data Analysis
Coding and cross-case analysis of Phase 1 transcripts against the rubric: Position 1/2/3 marks for every criterion-reach, the RQ1 judgment signatures by cell, and the RQ2 propagation signatures at the boundary handoff. Frontline analyzed first so the criteria shifts are documented before propagation is traced. Builds as interviews complete.
Cross-case Rubric-coded Frontline first
Planned
ANALYSIS
Framework Refinement → Paper 2
Where Study 1 findings revise the framework: which predictions held by cell, where the protection from consequence exposure ran out, whether the criterion survived or dropped out at the boundary handoff, and what the thin Veteran AI-light cell says about the population. Feeds the refined PSF account that becomes Paper 2.
Revises framework Feeds Paper 2
Planned
04 Study 2 Fieldwork · Multi-Domain 2 items
ADJACENT DOMAIN
AI Diagnostics in India
Research landscape across radiology, pathology, and clinical laboratory AI in Indian diagnostic centers. Tracks PSF-relevant open questions (outcome-measurement gaps, deskilling and automation bias, feedback-loop contamination, dataset bias as invisible accuracy gap, regulatory scaffolding). A regulated, life-critical, "most different" case for testing PSF outside the software mainline.
7 PSF questions "Most different" case With Neeti Gupta
Active · research interest
FIELDWORK · DESIGN
Study 2: Multi-Domain Design
Phase 2 fieldwork carrying the refined framework into domains beyond software, to test whether the mechanism is portable or software-bound. Domain selection, the comparative logic, and how the Study 1 instruments adapt to a setting where the accountable criterion lands materially. Grounds the multi-domain Paper 3.
Multi-domain Portability test Feeds Paper 3
Planned
05 Papers & Dissertation 4 items
PAPER 1 · THEORY
Theoretical Framework Paper
The framework paper that specifies the mechanism: comparative legibility, proxy elevation and evaluator transformation, the three dimensions of evaluative capacity, and the three-level erosion. Names the three research directions Study 1 then tests. Targeted at an Organization Science Perspectives outlet.
Mechanism 3 directions ETA Jun 2026
Active · drafting
PAPER 2 · REFINED PSF
Refined PSF Paper
The empirical paper in the dissertation sequence. Takes Study 1 findings from software development and the framework refinement they produce, and presents the revised account of where consequence exposure protects, where it runs out, and how elevated metrics propagate through the typification loop.
Study 1 grounded ETA Mar 2027
Planned
PAPER 3 · MULTI-DOMAIN
Multi-Domain Fieldwork Paper
PSF grounded across domains, drawing on Study 2. Tests whether the mechanism holds beyond software, including where the accountable criterion lands materially, and settles how far the framework's claims are mechanism-level rather than domain-bound.
Study 2 grounded Portability ETA Dec 2027
Planned
DISSERTATION
Dissertation
The integrating document. Writing runs through Year 3, drawing the framework paper and the two empirical studies into a single account of organizational evaluative capacity under transformative AI engagement, then polish and submission.
Integrates 3 papers ETA Jul 2028
Planned