Independent Human–AI Research Validator (Qualitative + Technical)

We are seeking one independent research validator for an existing 17,600-word scholarly manuscript, “The Model Knew Better: Sycophantic Falsification Failure in AI-Assisted Trading Research.” This is independent methodological validation—not ghostwriting, general academic writing, or software development. PROJECT The manuscript analyzes a prospectively logged 76-day human–AI technical research project involving prediction-market automation. The retained record includes 104 archived AI sessions, 65,897 messages, 20 focal strategy/experiment episodes, 18 replay-validity audits, 13 paper/live drift episodes, and bounded technical artifacts. The engagement has two related parts: 1. QUALITATIVE RESEARCH CODING / CONTENT ANALYSIS (NOT PROGRAMMING) Using a supplied codebook and structured spreadsheet, independently classify 20 bounded, deidentified human–AI interaction episode packets. You will not see the author’s final classifications before completing your own. Preserve your initial classifications, document uncertainty and disagreements, and participate in one reconciliation discussion. We may report percentage agreement and an appropriate inter-rater/inter-coder reliability statistic, depending on the final codebook structure. 2. TECHNICAL-METHODOLOGY REVIEW Independently assess approximately five bounded research-result reversals involving issues such as accounting scope, replay validity, hindsight/selection, correlated exposure, and execution/depth assumptions. Review supplied scripts, formulas, tables, result files, or bounded source data where useful. Determine whether each stated defect and correction are supported, whether before/after numbers are internally consistent, whether conclusions follow, and what caveats are needed. This is a methodological audit. You will not rebuild the trading system or reproduce the full historical dataset. No passwords, wallet keys, credentials, private infrastructure access, or unrestricted raw archive will be provided. DELIVERABLES • Completed independent qualitative research coding/content-analysis spreadsheet for 20 episodes, with uncertainty notes and preserved pre-reconciliation classifications • Short agreement/disagreement summary and one reconciliation session • Written assessment of approximately five technical reversal cases • Concise 3–6 page independent validation memo covering qualifications, materials, procedure, independence safeguards, agreement results, technical findings, and limitations IDEAL BACKGROUND Graduate-level research experience is strongly preferred. Relevant fields include HCI/human–AI interaction, qualitative or mixed methods, information systems, communication, STS, computational social science, and research methodology. You should have serious experience with qualitative content analysis, thematic analysis, and/or inter-rater/inter-coder reliability, plus enough technical/data literacy to interpret Python-generated outputs, replay/backtest summaries, tabular results, simple probability/expected-value reasoning, accounting logic, data-selection or hindsight issues, execution/fill assumptions, and methodological validity. Deep quantitative-finance or prediction-market expertise is not required. Please do not apply if your background is limited to generic academic writing, editing, programming, software development, or trading consulting without qualitative research-methods experience. We are not seeking a manuscript rewrite, guaranteed publication, or predetermined conclusions. INDEPENDENCE You must have had no involvement in the original project and must be comfortable concluding that the author’s proposed interpretation is unsupported. Initial classifications must remain independent; disagreements and uncertainty must be documented rather than silently harmonized. SCOPE, BUDGET, AND AUTHORSHIP Expected workload: approximately 20–30 total hours. Fixed-price target: $500–$750. We will begin with a paid $75–$100 pilot covering the codebook, four episode packets, and one small technical case. The remaining assignment will proceed only after the pilot is approved. This engagement is initially for paid independent methodological validation. Scholarly acknowledgement will be offered with the researcher’s permission. If the researcher later makes a substantial intellectual contribution to study methodology, interpretation, and manuscript revision, authorship may be discussed separately under normal scholarly authorship criteria. When applying, answer all screening questions and provide evidence of substantive research-methods work.