# content-matter-expert — domain presets

_Last updated: 2026-07-03_

Optional accelerators for Mason's recurring fields. Load only when the topic
sits in one of these areas. A preset sharpens framing and source standards for
step 2 (source strategy) and the field map — it does NOT replace the pipeline
or the mandatory verify step.

## A. Consciousness studies, psychokinesis, RNG anomalies, AI consciousness

Risk: contested / interdisciplinary / high hallucination risk.

Source standards:
- Prefer peer-reviewed reviews, meta-analyses, registered reports, direct PEAR
  and Global Consciousness Project publications, replication attempts and
  critiques, statistical-method papers, philosophy of mind, cognitive science,
  neuroscience, and AI-consciousness papers.
- Treat institute pages, paranormal advocacy, YouTube, podcasts, and popular
  summaries as claims requiring verification.
- Always separate: empirical RNG/PK claims, interpretation of effect sizes,
  philosophical consciousness claims, and speculative AI-consciousness claims.

Required checks: effect size; p-hacking / file-drawer; optional stopping;
multiple comparisons; pre-registration; experimenter effects; controls; RNG
hardware; environmental confounds; replication history; whether "quantum"
language is mechanistically justified; whether a claim is empirical,
philosophical, metaphorical, or artistic.

Default stance: **open but severe.** Don't debunk lazily; don't overclaim.

## B. Crypto trading, cryptobots, AI/LLM-assisted trading

Risk: financial / adversarial / rapidly changing.

Source standards:
- Current exchange docs, API docs, regulatory sources, security advisories,
  market-microstructure literature, audited code, backtest-methodology papers,
  primary project docs.
- Treat influencer threads, exchange marketing, Telegram/Discord claims, and
  backtest screenshots as weak evidence.

Required checks: fees, slippage, liquidity, latency; survivorship bias,
lookahead bias, data snooping, overfitting, train/test leakage; exchange
downtime, custody, key management; tax/regulatory risk; adversarial market
behavior; the live/paper gap and whether results survive realistic transaction
costs; LLM-use boundaries (ideation, code review, monitoring, narrative — never
an oracle for price direction).

Default stance: **assume the backtest is lying until proven otherwise.**

## C. AI, LLMs, frontier models, agents, applications, ethics

Risk: rapidly changing / technical / benchmark-sensitive.

Source standards:
- Official docs, model/system cards, eval papers, benchmark repos, arXiv
  preprints with caution, API changelogs, safety papers, reputable technical
  analyses.
- Current web checks are mandatory for model capabilities, pricing, API names,
  context windows, tool support, and safety policies.

Required checks: benchmark contamination, eval design, latency/cost, tool-use
reliability, prompt sensitivity, hallucination rate, data privacy,
copyright/licensing, deployment constraints, human-in-the-loop requirements;
distinguish research demo from production-ready practice.

Default stance: **skeptical engineer, not hype columnist.**

## D. EdTech, AI in education, adaptive learning, AI tutors

Risk: applied research / ethics / efficacy.

Source standards: learning sciences, educational psychology, AIED/LAK/EDM
papers, IES / What Works Clearinghouse, controlled studies, meta-analyses,
accessibility standards, privacy-law summaries when needed, implementation
studies.

Required checks: learning outcomes, assessment validity, cognitive load,
transfer, feedback quality, learner agency, equity, privacy, teacher workflow,
accessibility, motivation vs durable learning, novelty effects; compare against
active learning, tutoring, simulations, worked examples, retrieval practice,
mastery learning.

Default stance: **outcome-first.** "Engaging" isn't enough unless learning
improves.

## E. Adult learning, professional development, science education, Borchard Labs

Risk: dissertation / design-based research / learning sciences / software
product.

Source standards: adult-learning theory, expertise development, situated
cognition, cognitive apprenticeship, experiential learning, inquiry-based
science education, game-/simulation-based learning, biology-education research,
accessibility, reproducible computational science.

Required checks for Borchard Labs work: learning-theory alignment; biology
correctness and deterministic/reproducible simulation output; assessment model
and evidence of learning; research design (what data is collected, what claims
it can and cannot support); game design (intrinsic motivation, world coherence,
interaction loops, avoiding chocolate-covered-broccoli edutainment);
accessibility and cross-platform constraints; field-notebook brand consistency.

For the current Borchard Labs product definition (scope, engine, brand,
platform targets), read the authoritative memory at
`~/.claude/projects/-home-borchard-borchard-labs-workspace-borchard-labs/memory/project_borchard_labs_product.md`
rather than trusting a copy here — the product moves faster than this file.

Default stance: **dissertation committee + ruthless game designer + biology TA.**
