# Research That Scales > [!summary] Recommendation > Use this as a systems-design handbook for research operations, especially when a team needs to connect strategy, recruitment, knowledge, support, tooling, ethics, budgets, prioritization, and talent. Towsey's planning questions and lifecycle models are stronger than the book's claims of broad effectiveness. Treat the large-company stories, self-selected polls, vendor surveys, AI snapshot, and executive-facing metrics as hypotheses to test locally. Add explicit decision rights, equity measures, participant outcomes, and decommissioning rules before adopting the frameworks. **Rating: 4/5.** ## Why this book belongs in this vault The book connects product discovery, DevRel, technical communication, knowledge management, service design, privacy, accessibility, and organizational strategy. It belongs beside [[3. Resources/Permanent Notes/A Working Strategy Is Measured by What It Lets a Team Refuse]], [[3. Resources/Permanent Notes/A Metric Is Vanity Only Against Your Own Outcome Data]], [[3. Resources/Permanent Notes/A Stale Backlog Creates Shadow Systems]], [[3. Resources/Permanent Notes/An Organization Cannot Expose Data to a New System It Cannot Locate Internally]], [[3. Resources/Permanent Notes/Retrieval Quality Is Set by the Unit of Knowledge Not the Retrieval Algorithm]], [[3. Resources/Permanent Notes/Training as a System Compounds Where a Session Decays]], and [[3. Resources/Permanent Notes/Your System Is the Practice]]. Its central contribution is a shift from scaling the amount of research to designing the conditions under which research remains usable, ethical, trusted, and sustainable as demand and variety change. Its central limit is evidential: experienced practice and selected cases show plausibility, but they rarely isolate causes or prove transfer across organizations. --- ## Prompt 0 — retrieval, provenance, and source map ### Source record | Field | Result | |---|---| | Supplied artifact | Local EPUB | | Integrity | Every archived file passed ZIP integrity testing; all EPUB XHTML and OPF files passed XML validation | | SHA-256 | `a3b2594dfd1c691662728a61e22fefb396d4ac0e59f900f792f6dfe584a37336` | | Metadata | *Research That Scales: The Research Operations Handbook*, Kate Towsey, Rosenfeld Media, 2024 | | Edition identity | First edition; ebook ISBN `9781959029205`; print ISBN `9781959029229` | | Official check | The [publisher record](https://rosenfeldmedia.com/books/research-that-scales/) confirms the author, September 2024 publication, ISBNs, 336-page print edition, audience, and twelve-chapter contents | | Extracted size | 93,911 plain-text words; 100,194 structure-preserving Markdown words | | Coverage | Front matter; introduction; chapters 1–12; coda; 146 chapter notes; 54 images; index; all substantive sections read in full | | Readwise Reader | Exact title/author search on 2026-08-27 returned no matching document | | Readwise highlights | Exact title/author search on 2026-08-27 returned no matching highlight record | | Reader layer | No reader notes, colors, timestamps, locations, or annotations supplied; reader highlight count `0` | | Canonical specification | `/Users/courtneyrobertson/Documents/2nd Brain/6. Templates/Books/Work the Book.md`, spec v2 | | Working note | `4. Archives/Book reviews/Research That Scales.md` | ### Layer 1 substitution The complete EPUB substitutes for Readwise as Layer 1. It preserves the full text, spine order, chapter notes, print page markers, navigation, captions, and image alternatives. It does not preserve reader emphasis. Every excerpt and hotspot in this note is **system-selected**. Nothing selected here should be read as Courtney's agreement, interest, prior belief, or reason for reading. No separate highlights note or session-record file was created. The explicit one-file contract controls this run, so the ingestion record, analysis, teaching pack, actions, visuals, and session record all live here. ### Stable anchor scheme - `EPUB-FRONT` — How to Use This Book, FAQ, Contents at a Glance, and Elizabeth F. Churchill's foreword - `EPUB-INTRO` — Introduction, “How to Eat an Elephant” - `EPUB-C01` — Research Does Not Scale—Systems Do, print pages 1–22 - `EPUB-C02` — Lost and Won on Strategy, pages 23–44 - `EPUB-C03` — From Strategy to Operational, pages 45–67 - `EPUB-C04` — Planning Realistic ResearchOps, pages 69–94 - `EPUB-C05` — People to Take Part in Research, pages 95–124 - `EPUB-C06` — Long Live Research Knowledge, pages 125–149 - `EPUB-C07` — Seamless Support, pages 151–176 - `EPUB-C08` — Tactical Tooling, pages 177–202 - `EPUB-C09` — Respect in Research, pages 203–230 - `EPUB-C10` — Money and Metrics, pages 231–250 - `EPUB-C11` — Getting Priorities Straight, pages 251–272 - `EPUB-C12` — Making the Most of Research Talent, pages 273–290 - `EPUB-CODA` — Time to Think, page 291 Append `#secN`, `#figN-N`, or `#pg_N` to cite an XHTML section, figure, or print-page marker. ### Highlight index Reader highlights: **0**. The following fourteen source passages form a representative, system-selected index for analysis. They are not a substitute for reader highlights. | ID | Chapter | Location | System-selected passage | Repeated concepts | |---|---|---|---|---| | EPUB-INTRO#sec1 | Introduction | opening | ResearchOps recombines existing systems and people skills | profession, skills, systems | | EPUB-C01#sec3 | Ch. 1 | pp. 4–6 | Scale concerns behavior under changing demand, not raw size | scale, demand, value | | EPUB-C02#sec6 | Ch. 2 | pp. 31–34 | Research strategy chooses what the practice will and will not do | strategy, priorities, refusal | | EPUB-C03#sec4 | Ch. 3 | pp. 52–54 | An operating model describes how actors and systems exchange value | design, model, value | | EPUB-C04#sec8 | Ch. 4 | pp. 87–93 | The Planning Matrix crosses eight elements with three maturity phases | planning, lifecycle, interdependence | | EPUB-C05#sec10 | Ch. 5 | pp. 110–123 | Recruitment is a ten-stage participant journey | participants, consent, support | | EPUB-C06#sec8 | Ch. 6 | pp. 139–147 | RKM combines communities of practice with governed knowledge services | knowledge, trust, findability | | EPUB-C07#sec13 | Ch. 7 | pp. 161–176 | Support needs a strategy, journeys, channels, and operating choices | support, capacity, access | | EPUB-C08#sec4 | Ch. 8 | pp. 184–197 | A tooling blueprint maps people, data, integrations, and maintenance | tools, data, culture | | EPUB-C09#sec1 | Ch. 9 | pp. 204–206 | Ethics, privacy, and governance form a respect-in-research trifecta | consent, care, risk | | EPUB-C10#sec7 | Ch. 10 | pp. 241–247 | Metrics should make operations visible and lead to assigned action | measures, budget, action | | EPUB-C11#sec2 | Ch. 11 | pp. 254–258 | Unspoken contracts conceal decision rights and priority conflicts | power, demand, ownership | | EPUB-C12#sec2 | Ch. 12 | pp. 276–281 | Competency frameworks connect hiring, growth, and performance | talent, equity, skills | | EPUB-CODA | Coda | p. 291 | Operating systems need protected time for whole-system review | reflection, cadence, maintenance | **Density and recurrence:** Chapters 5–9 contain the densest operational detail. `Strategy`, `system`, `value`, `culture`, `support`, `people`, `research`, `operations`, `trust`, `data`, and `metrics` recur across the book. Transit, city, journey, map, ecosystem, and lifecycle metaphors repeatedly turn organizational work into navigable infrastructure. ### Cluster map | Cluster | Span | Main topics | Rationale | Representative source IDs | |---|---|---|---|---| | A — Purpose and system design | Introduction; Ch. 1–4 | scale, strategy, operating models, eight elements, maturity | These chapters form the required conceptual chain | EPUB-C01#sec3; EPUB-C02#sec6; EPUB-C03#sec4; EPUB-C04#sec8 | | B — Research infrastructure | Ch. 5–8 | recruitment, knowledge, support, tools, AI | These chapters design the routes through which research is done and reused | EPUB-C05#sec10; EPUB-C06#sec8; EPUB-C07#sec13; EPUB-C08#sec4 | | C — Governance and allocation | Ch. 9–11 | ethics, privacy, budgets, metrics, priorities, program management | These chapters constrain risk and allocate money, attention, and authority | EPUB-C09#sec1; EPUB-C10#sec7; EPUB-C11#sec2 | | D — People and renewal | Ch. 12; coda | hiring, competency, onboarding, growth, reflection | The final movement treats people and review time as system maintenance | EPUB-C12#sec2; EPUB-C12#sec9; EPUB-CODA | ### Source defects, corrections, and limits 1. Whole-book Pandoc conversion emitted unresolved internal-reference warnings even though the original chapter-local footnote targets exist. The original XHTML, not the conversion, controls quotations and notes. 2. Chapter 6 says 40 reports per month for three years yields 1,140 reports. The arithmetic is 1,440 (`EPUB-C06#sec2`). 3. Chapter 6 footnote 16 prints *The Tactic Dimension*; the Polanyi work discussed in the text is *The Tacit Dimension*. 4. Chapter 3 describes a daily 1% compound change as a “37.78% shift.” The factor is about 37.78, equivalent to roughly 3,678% growth, before debating whether the analogy fits organizations (`EPUB-C03#sec1`). 5. Chapter 4 gives the gravitational constant as `6.67408 m/s²`, omitting the exponent and correct dimensions. The analogy does not determine the ResearchOps framework (`EPUB-C04#ch4`). 6. The EPUB contains 45 nondecorative images with nonempty alt text and 9 decorative images with empty alt text and `role="presentation"`. Some complex diagrams receive summaries rather than full structural equivalents. 7. The package claims EPUB Accessibility 1.1 conformance at WCAG 2.2 Level A and says the publication was generally produced toward AA, but it discloses no check for language shifts or color contrast inside images. This review did not run an assistive-technology test. --- ## Prompt 1 — reading contract, priors, and open questions > [!warning] 🔶 Inferred reader inputs > No purpose, prior-belief inventory, constraints beyond one-file output, existing-knowledge statement, or DEIB goal accompanied the EPUB. I infer a practical purpose: test the book as an operating guide for research-heavy product, DevRel, technical communication, and knowledge work. I select analytical depth and use participant dignity, disability access, cross-cultural transfer, power, and labor sustainability as the inclusion lens. These are system assumptions, not reader beliefs. A different purpose could change the action and relevance judgments. ### Reading contract - **Why read:** Decide which ResearchOps practices can improve research systems and which need stronger evidence or local adaptation. - **What to understand:** How strategy becomes infrastructure; how the eight operational elements connect; what makes the system ethical, findable, supportable, measurable, and sustainable. - **Intended use:** A vault reference, a critique of operations frameworks, a source of reversible experiments, and a teaching resource. - **Success criteria:** Recover the book's full argument; trace claims to chapters; separate author experience from independent evidence; identify failure modes; produce bounded actions. - **Evidence that would change the conclusion:** Comparative studies of operating models, participant-outcome data, audited cost and quality measures, failed implementations, and examples from small, public, nonprofit, global-majority, and regulated settings. - **What this source cannot establish:** Universal team structures, causal business impact, legal compliance in a specific jurisdiction, current AI capabilities, or the best vendor stack. - **Constraints:** One consolidated file; no reader highlights; no access to organizations' internal results; current legal and product facts can drift. ### Prior-belief and incentive guardrails Reader-supplied beliefs and motives are unknown. The system begins with these provisional beliefs: - Strategy matters when it changes refusal, allocation, and maintenance decisions. - Tools encode behavior but cannot repair unclear ownership or missing policy. - A knowledge repository without stewardship becomes an accumulation problem. - Participant recruitment is an ethical relationship, not just scheduling. - Central standards can reduce friction and also concentrate power. - Metrics need denominators, decision links, and outcome checks. - Support and maintenance labor become invisible when teams report only new delivery. - AI needs data, consent, provenance, review, and accountability controls beyond “human in the loop.” - Accessibility must be built into standard paths, not reserved for exceptions. - Systems should include retirement, appeal, incident, and correction routes. The inferred professional context may bias the analysis toward Towsey's systems argument because it validates connective, documentation, and knowledge work as core infrastructure. Prompt 7 tests that bias by looking for hidden labor, centralized power, proxy metrics, exclusion through standardization, and selected success stories. What the reader wants to be true or may resist remains **unknown**. No personal, social, ideological, or financial motive is inferred. ### Open questions inventory #### Definitions and mechanisms 1. What observable behavior distinguishes a scalable research system from a larger one? 2. Which parts of a research operating system must be standardized, and which must remain variable? 3. When does culture cause system adoption, and when does system design cause culture? 4. Does the eight-element model describe reality or impose a useful planning lens? 5. How do tacit knowledge and formal knowledge services interact without becoming a false binary? #### Evidence and credibility 6. Which practices have outcome evidence independent of Towsey's experience? 7. What failed ResearchOps systems would test the success-case pattern? 8. Does the 2024 LinkedIn poll say anything beyond its self-selected respondents? 9. Which dashboards predict research quality, participant welfare, or decision quality? 10. How much of the book's advice depends on large-company authority and funding? #### Application and failure modes 11. What is the smallest useful operating system for a solo or volunteer research practice? 12. When does a planning matrix clarify dependencies, and when does it hide recursive work? 13. How should a team retire a tool, taxonomy, service, or metric? 14. How can support protect scarce expertise without blocking urgent safety or access needs? 15. When does a study tracker become surveillance or a stale shadow system? 16. How should teams protect exploratory work when priorities follow visible business goals? #### Power, DEIB, and multicultural transfer 17. Who defines participant cohorts, taxonomies, quality, and value? 18. Who can appeal a recruitment, ethics, privacy, funding, or priority decision? 19. Does inclusive recruitment change who sets the research questions and interprets results? 20. How do compensation, device access, language, time zones, disability, and legal status change participation? 21. Which claims depend on UK, US, Australian, or large-technology-company assumptions? 22. How should Indigenous data sovereignty and community governance change the model? #### Accessibility and unresolved questions 23. Are standard research paths accessible before an accommodation request appears? 24. How should neurodivergent researchers and participants shape timing, consent, support, and synthesis? 25. Can a concise image alternative convey the operational relationships in a complex matrix? 26. What protections are needed when AI transcribes, translates, classifies, or summarizes sensitive research? 27. What should count as evidence that an operating system improved rather than merely became more visible? ### Priors seed | Claim | Initial belief | Prior confidence | Reason | Evidence needed | |---|---|---:|---|---| | P-01 Systems, not raw volume, determine scale | Agree | 75% | Capacity and coordination failures are system-shaped | Defined demand, quality, cost, and failure measures across cases | | P-02 Research and operations strategies must align | Agree | 80% | What/how alignment is logically strong | Cases comparing aligned and unaligned practices | | P-03 Eight elements cover ResearchOps well | Uncertain | 55% | Plausible map, but category completeness is hard to prove | Competing taxonomies and gap analysis | | P-04 Planning across lifecycle phases improves sustainability | Agree | 75% | Maintenance is a known source of system failure | Longitudinal outcome and retirement data | | P-05 Standardized recruitment improves access and quality | Mixed | 60% | Standards reduce omission but can encode blind spots | Completion, quality, burden, and equity outcomes | | P-06 RKM needs stewardship beyond a repository | Agree | 85% | Findability and trust depend on governance | Comparative repository/service evidence | | P-07 Designed support reduces hidden labor and friction | Agree | 75% | Channel and capacity design affects access | Task-success, wait-time, quality, and burden data | | P-08 Tools shape research culture | Agree | 70% | Defaults alter behavior | Before/after use and quality measures with alternatives | | P-09 Ethics, privacy, and governance must be integrated | Agree | 90% | These risks share data and decision paths | Incident, comprehension, withdrawal, and remediation outcomes | | P-10 Metrics make operations more governable | Mixed | 65% | Visibility helps, but proxies invite gaming | Measures tied to decisions and participant/research outcomes | | P-11 Explicit priority contracts reduce conflict | Agree | 75% | Decision-right ambiguity creates recurring failure | Conflict, cycle-time, appeal, and satisfaction outcomes | | P-12 Competency frameworks improve talent systems | Mixed | 60% | Clarity helps; scoring can preserve bias | Hiring, promotion, retention, pay, and appeal audits | ### Drift assumptions - AI capabilities, vendor products, synthetic-participant claims, model policies, and prices can change within months. - Privacy, employment, research, accessibility, and AI regulation differ by jurisdiction and continue to change. - The ResearchOps labor market, organizational structures, and professional vocabulary can change within a year. - Named platforms, surveys, and workforce figures may become stale. - The book's claims are fixed. The evaluative context around them is live. - Lower-drift ideas include aligning strategy with operations, governing knowledge lifecycles, keeping consent ongoing, making decision rights explicit, and planning maintenance before launch. --- ## Prompt 2 — inspectional reading ### Central project Towsey asks how organizations can increase research reach and value without treating research like mass production. Her answer is to design a research operating system: a connected set of strategy, people, practices, services, tools, data rules, measures, and cultural signals that can respond to changing demand. She teaches through definitions, planning frameworks, service-design models, first-person cases, third-party cases, checklists, and metaphors. The intended audience is research leaders, researchers, ResearchOps practitioners, and people in adjacent fields who support human-centered research. The main rhetorical modes are **teach**, **explain**, **persuade**, **advocate**, **synthesize**, and **warn**. ### Book map and coverage | Section | Purpose and central topic | Relation to argument | Available evidence | Coverage | Mapping confidence | |---|---|---|---|---|---:| | Front matter | Audience, sequence, myths, endorsement | Defines use and stakes | Author scope; Google foreword case | Read | High | | Introduction | Professional history and adaptable framing | Establishes author position and method | First-person history; named organizations | Read | High | | Ch. 1 | Define scale and research operating system | Foundational definition | Analogies, cases, cited definitions | Read | High | | Ch. 2 | Define research and ResearchOps strategy | Selects purpose and priorities | Poll, trade literature, author model | Read | High | | Ch. 3 | Turn strategy into operating models | Bridge from intent to system design | H&M case, GDS case, model library | Read | High | | Ch. 4 | Eight elements and maturity phases | Full planning scaffold | Worked examples and matrix | Read | High | | Ch. 5 | Participant recruitment | First operational element | Author case, vendor survey, workflow | Read | High | | Ch. 6 | Research knowledge management | Preserves and reuses research value | Knowledge theory, Atlassian case | Read | High | | Ch. 7 | Research support | Keeps the system usable | Service models, author experience | Read | High | | Ch. 8 | Tooling and AI | Encodes workflow, data, and culture | Tool case, survey, procurement model | Read | High | | Ch. 9 | Ethics, privacy, data governance | Constrains harm and protects people | Facebook case, legal/ethics sources, author cases | Read | High | | Ch. 10 | Budgets and metrics | Funds and observes operations | Budget practice, sample reports | Read | High | | Ch. 11 | Prioritization and program management | Allocates attention and decision power | GDS, NASA, author cases | Read | High | | Ch. 12 | People and skills | Builds and sustains research talent | HR frameworks, company examples | Read | High | | Coda | Protected reflection | Adds review cadence to the operating system | Author's quarterly practice | Read | High | ### Provisional 150-word summary > [!abstract] Provisional summary > *Research That Scales* argues that research itself cannot be mass-produced without losing the judgment and variation that make it useful; organizations must instead scale the systems around it. Towsey links a research-specific strategy to an operations strategy, then turns that alignment into operating models and a planning matrix spanning eight interdependent elements: recruitment, knowledge, support, tooling, respect, money and metrics, prioritization, and talent. Her strongest practical idea is lifecycle thinking: build or buy, specialize and improve, then keep the system working. The book also treats culture, trust, consent, maintenance, and hidden labor as operating concerns rather than side issues. Its frameworks are concrete and adaptable, but much of the support comes from Towsey's experience, selected large-company cases, trade sources, vendor research, and analogy. The book works best as a disciplined question set for local design and testing, not as proof that one operating model, metric, taxonomy, or team structure will work everywhere. ### Reading hotspots | Hotspot | Why it matters | |---|---| | Ch. 1, `#sec3–#sec11` | Defines scale, operating system, decision-tree logistics, and culture | | Ch. 2, `#sec3–#sec13` | Defines strategy, value, trust, and the what/how split | | Ch. 3, `#sec3–#sec14` | Supplies the main system-design and operating-model choices | | Ch. 4, `#sec1–#sec8` | Introduces the eight elements, maturity phases, and Planning Matrix | | Ch. 5, `#sec10–#sec21` | Connects recruitment workflow, consent, deletion, support, and inclusion | | Ch. 6, `#sec2–#sec11` | Explains why repositories fail and how trust and stewardship work | | Ch. 8, `#sec4–#sec11` | Maps tool lifecycles, data movement, selection, and dated AI guidance | | Ch. 9, `#sec1–#sec13` | Offers the clearest ethics, privacy, governance, and consent argument | | Ch. 10, `#sec7–#sec10` | Shows the value and risk of executive-facing metrics | | Ch. 11, `#sec2–#sec12` | Makes priority power, contracts, decision rules, and trackers explicit | | Ch. 12, `#sec2–#sec9` | Applies operations to competency, hiring, accessibility, and onboarding | | Coda | Adds a recurring whole-system review practice and exposes schedule privilege | ### Guiding questions for deeper reading 1. What measured thresholds make a system scalable rather than merely larger? 2. What evidence supports the causal role assigned to culture? 3. Does the research/ResearchOps strategy split clarify ownership or create two documents for one decision system? 4. Which operating-model pairs describe spectra rather than choices? 5. What is missing from the eight elements? 6. How do standardization and inclusion constrain each other? 7. Who governs participant taxonomies and knowledge taxonomies? 8. What outcome makes a repository, support channel, or tool successful? 9. Which ethical decisions remain open when a checklist is complete? 10. Which measures change decisions rather than improve presentation? 11. Who owns priority appeals and protected exploratory capacity? 12. Which talent practices change outcomes instead of polishing process? ### Reading strategy **Recommendation: full analytical read, completed.** The book's parts form a dependency chain, and the ethics, knowledge, support, metrics, and talent chapters change how the opening systems argument should be judged. A selective read would miss those cross-element effects. The recommendation would change to syntopical comparison if the next decision required choosing among ResearchOps taxonomies or proving effectiveness rather than designing a local pilot. --- ## Prompt 3 — classification and terminology ### Classification - **Domain:** Business operations and organizational design - **Subdomains:** ResearchOps, user research, service design, knowledge management, governance, talent operations - **Genre:** Practitioner handbook with frameworks, cases, checklists, and professional argument - **Argument:** Definitional, methodological, practical, prescriptive, and partly causal - **Primary method:** Experience-based synthesis joined to third-party cases and operating models - **Intended use:** Diagnose, plan, build, maintain, and communicate a research operating system - **Comparison traditions:** Operations management, sociotechnical systems, systems thinking, service design, knowledge management, DesignOps, DevOps, research ethics, program management ### Controlled vocabulary Definitions in this table are paraphrases unless quotation marks appear. | Term | Author's definition | Competing definition | Operational definition | Equivocation risk | Example | Counterexample | Evidence | |---|---|---|---|---|---|---|---| | Scale | Predictable, proportional, measured, efficient, necessary, and valuable change under demand | Growth in size or reach | Capacity changes while quality, cost, access, and harm stay within set limits | Can slide between size, efficiency, resilience, and value | A recruitment service handles more varied studies without rising failure | More studies with longer waits and poorer representation | EPUB-C01#sec1–#sec6 | | Scalable system | A system that adjusts as demand changes | A system that can serve a large maximum load | A monitored system with tested demand ranges and failure thresholds | “Efficient” and “sustainable” can become assumed virtues | Support routes by risk and channel | One expert answers every request | EPUB-C01#sec3–#sec5 | | Research | Broad knowledge-gathering or learning activity; rigor depends on strategy and context | Systematic inquiry meeting disciplinary standards | State purpose, method, participants, evidence, and quality threshold | Broad use can erase the line between conversation and study | Moderated study with declared method | Casual customer chat reported as representative | EPUB-C01#ch1 note | | ResearchOps | Systems and people work that enables research action, value, and scale | Administrative support for researchers | Design and stewardship of research infrastructure, rules, services, and relationships | Can absorb HR, legal, IT, finance, and research leadership | Governed recruitment plus consent and deletion | Tool procurement alone | EPUB-FRONT; EPUB-INTRO | | Research operating system | Interconnected, shared, trusted assets and actions for research | Software operating system or tool stack | People, services, rules, data, tools, and feedback loops with owners | Metaphor may imply more unity than organizations possess | Connected intake, recruitment, storage, support, and review | A folder of templates and vendor links | EPUB-C01#sec7–#sec11 | | Research culture | Beliefs, biases, attitudes, and values shaping research | Observable norms and incentives | Patterns of decisions, participation, trust, refusal, and resource allocation | Risks explaining every outcome after the fact | Leaders protect consent despite deadline pressure | Values statement with coercive practice | EPUB-C01#sec10–#sec11 | | Research strategy | What the practice will and will not do to deliver organizational value | A roadmap, vision, or strategic research agenda | A small set of prioritized choices, exclusions, measures, and review rules | Can center executive perception instead of truth or welfare | Refuse low-priority studies to protect a goal | List every desirable activity | EPUB-C02#sec4–#sec6 | | ResearchOps strategy | How operations will enable research priorities and control cost | Operations backlog or tool plan | Operating choices, owners, resources, measures, and maintenance tied to research strategy | “How” can contain hidden “what” decisions | Fund an inclusive panel because strategy needs underserved users | Buy a repository without a knowledge goal | EPUB-C02#sec6–#sec7 | | Operating model | How systems and actors create and exchange value | Org chart or process diagram | Named service, control, demand, and automation choices with owners and exceptions | Paired models can look binary | Hybrid support by risk | “Self-service” with undocumented human cleanup | EPUB-C03#sec4–#sec13 | | Eight elements | Recruitment; knowledge; support; tools; respect; money/metrics; program management; people/skills | ResearchOps Community's eight pillars | A checklist for dependency discovery, not isolated departments | “Elements” may imply completeness or separability | Tool plan checks privacy, budget, support, and training | Tool selected on features alone | EPUB-C04#sec1–#sec3 | | Operational maturity | Build/buy; specialize/improve; keep the lights on | Capability maturity levels | Lifecycle states with entry, health, incident, and exit criteria | “Maturity” can imply one-way progress | Pilot, improve, maintain, retire | Perpetual pilot with no owner | EPUB-C04#sec4–#sec8 | | KTLO | Ongoing work that keeps a live system reliable | Minimal maintenance | Scheduled ownership for routine, corrective, security, access, and support work | Can hide labor under a maintenance label | Renew consent system and audit deletion | Ignore old workflows after launch | EPUB-C04#sec7 | | Participant taxonomy | Controlled grouping of commonly needed populations | Demographic segmentation | Governed, revisable cohorts linked to research purpose and access routes | Categories can encode institutional blind spots | Decision-makers versus end users | “Customers” as one undifferentiated cohort | EPUB-C05#sec1 | | RKM | Research knowledge management across creation, storage, access, trust, and use | A research repository | Lifecycle governance for research evidence and knowledge products | May blur records, synthesis, and organizational learning | Curated service plus community practice | Uploading reports without review | EPUB-C06#sec1–#sec12 | | Knowledge service | Centrally curated, consumer-facing knowledge support | Any knowledge platform | A governed service with submission, provenance, access, review, and retirement | “Service” can hide classification authority | Trusted library with stewards | Search across abandoned drafts | EPUB-C06#sec8–#sec11 | | Logistical support | Rule-based how-to help | Administrative assistance | Repeatable access and process help with escalation | “Rule-based” may still involve judgment | Account access and consent steps | Method choice for a sensitive study | EPUB-C07#sec2 | | Craft support | Contextual what-to-do research guidance | Training or mentoring | Expert judgment on method, quality, and interpretation | Gatekeeping may be called quality control | Coaching a study design | Password reset | EPUB-C07#sec3–#sec4 | | Research tooling blueprint | Service blueprint for research tools, people, workflows, and data | Tool inventory | End-to-end map of users, steps, data, integrations, support, risk, and exit | A map can be mistaken for tested interoperability | Consent data traced through systems | Vendor list by price | EPUB-C08#sec4–#sec5 | | Respect-in-research trifecta | Ethics, data governance, and data privacy with consent at the center | Compliance program | Joined controls for care, purpose, access, retention, deletion, and remedy | Legal, ethical, and procedural adequacy can collapse together | Ongoing consent with withdrawal and deletion | Signed form with unlimited reuse | EPUB-C09#sec1–#sec13 | | Operating metric | Number that reveals inputs, flow, outputs, quality, or experience | KPI or performance score | Defined measure with denominator, owner, decision, guardrail, and review date | Visibility can pose as value or causality | Wait time plus completion quality and access | Session count alone | EPUB-C10#sec7–#sec9 | | Collaboration model | Shared agreement for roles, commitments, decisions, and failure handling | Team charter | Versioned operating agreement with appeal and exit routes | Agreement may conceal unequal bargaining power | Researcher, funder, and manager define priority rights | Informal expectations | EPUB-C11#sec2–#sec5 | | Competency framework | Behaviors, responsibilities, and skills by role and level | Performance rubric | Co-designed, behavior-based expectations with calibration and appeal | Can turn contextual work into scores | Structured hiring criteria | Promotion by manager impression | EPUB-C12#sec2 | ### Contested language - **Scale** shifts among size, proportional response, resilience, efficiency, sustainability, and value. - **Value** moves between organizational perception, cost efficiency, decision support, and user benefit. - **Trust** refers to confidence in researchers, content, systems, and organizational intent. - **Culture** can name observed behavior, shared belief, leadership expectation, or a post-hoc explanation. - **Democratization** can mean expanded participation or transfer of research labor without authority and support. - **Inclusive** sometimes names a process intention and sometimes an outcome; the difference matters. - **Self-service** can hide expert labor, exception handling, and inaccessible standard paths. - **AI** groups different systems and tasks under one label; 2023 capability observations are not current product tests. - **Impact** can mean visibility, use, changed decisions, changed products, or improved participant/user outcomes. ### Category-error watchlist | Risk | Status | Where | Effect | |---|---|---|---| | Correlation treated as causation | Present | H&M, Google, tooling culture, team-brand stories | Cases support plausibility, not isolated effects | | Individual versus institutional behavior | Present | Strategy, trust, resilience, talent | Personal skill can be asked to repair missing authority or resources | | Description versus prescription | Present | Eight elements, maturity phases, support models | A useful map can become a mandatory sequence | | Identity versus behavior | Limited | Participant and talent categories | Demographic grouping can harden into assumptions | | Intention versus effect | Present | inclusion, engagement strategy, culture | Good intent does not establish access, equity, or trust | | Anecdote versus population evidence | Present | Author cases and selected company examples | Strong for mechanism discovery, weak for prevalence or effectiveness | | Legal versus moral claims | Explicitly resisted | Ch. 9 | This is a strength; local legal accuracy still needs review | | Technical capability versus practical outcome | Present | tools, APIs, AI, automation | Feature availability does not prove research quality or safe use | | Absence of evidence versus evidence of absence | Present risk | Missing failed cases and outcome data | The book's silence cannot prove failure is rare | | Theological or historical category error | Not applicable | — | The book makes no theological argument; history is secondary context | --- ## Prompt 4 — claims, dependencies, and falsifiers ### Claim-source labels - **EA:** explicit author claim - **IA:** implied author claim - **Q3P:** quoted or summarized third-party claim - **RI:** reader inference; none are used in the author-claim inventory - **UI:** unresolved interpretation ### Claims inventory | ID | Source | Claim | Type and scope | Testability and expected evidence | Non-vacuous falsifier | Key terms | Source | Question links | |---|---|---|---|---|---|---|---|---| | C-001 | EA | Research does not scale like manufacturing; systems around research scale | Definitional, broad | Cross-context demand/quality/cost data; partial cases appear | Research volume rises across contexts with stable quality and cost without system change | scale, system | EPUB-C01#sec3–#sec7 | Q1, Q2 | | C-002 | EA | A scalable system responds predictably and valuably as demand changes | Definitional, methodological | Measured capacity curves and failure thresholds; not supplied comparatively | Systems classed as scalable repeatedly fail stated quality, cost, or access thresholds | scalability, demand | EPUB-C01#sec3–#sec5 | Q1, Q27 | | C-003 | EA | A research operating system must be complete, commutable, and cultural | Evaluative, broad | Comparative system maps and outcomes; author cases appear | A disconnected system performs as well across complex journeys and adoption contexts | operating system, culture | EPUB-C01#sec7–#sec12 | Q3, Q11 | | C-004 | EA | Research strategy defines what to do; ResearchOps strategy defines how | Methodological, organizational | Strategy artifacts linked to allocation and outcomes; examples appear | Teams with no paired strategies match aligned teams on focus, value, and sustainability | strategy, operations | EPUB-C02#sec4–#sec7 | Q3, Q10 | | C-005 | EA | Research value should be judged at practice/system level, not attributed to one study's bottom-line effect | Evaluative, organizational | Portfolio measures and decision traces; logic appears | Reliable study-level causal attribution predicts organizational value better than portfolio measures | value, cost center | EPUB-C02#sec8–#sec12 | Q9, Q27 | | C-006 | EA | Strategic priorities become real only through changed operations | Causal, broad | Before/after cases with alternatives; selected cases appear | Priorities succeed repeatedly without material operating change | operating model | EPUB-C03#ch3–#sec4 | Q6, Q12 | | C-007 | EA | ResearchOps initiatives implicate all eight elements | Methodological, broad | Dependency audits across initiatives; worked examples appear | Multiple successful initiatives show no meaningful dependency on one or more elements | eight elements | EPUB-C04#sec1–#sec3 | Q4, Q5 | | C-008 | EA | Planning build/buy, improvement, and KTLO improves long-term sustainability | Practical, lifecycle | Longitudinal service-health and retirement data; no comparison appears | Lifecycle planning shows no reduction in failure, abandonment, or unplanned labor | maturity, KTLO | EPUB-C04#sec4–#sec8 | Q12, Q13 | | C-009 | EA | Recruitment should be standardized end to end while preserving cohort-specific paths | Practical, recruitment | Quality, cycle, burden, representation, and consent outcomes; workflow and survey appear | Standardization raises exclusion, burden, or poor matches without compensating gains | taxonomy, consent | EPUB-C05#sec1–#sec21 | Q6, Q17, Q20 | | C-010 | EA | RKM needs strategy, stewardship, community practice, and knowledge services—not a repository alone | Practical, knowledge | Findability, reuse, trust, decision, and maintenance evidence; theory and cases appear | Unstewarded repositories match governed services on trust, reuse, and retention | RKM, knowledge service | EPUB-C06#sec1–#sec12 | Q5, Q7, Q18 | | C-011 | EA | Research support must be designed as a system with bounded channels and capacity | Practical, support | Task success, wait time, quality, access, and labor data; examples appear | Ad hoc support matches designed services without greater hidden cost or inequity | support, SLA | EPUB-C07#sec1–#sec22 | Q14, Q23 | | C-012 | EA | Tools shape research culture and need full-lifecycle governance | Causal and practical | Use patterns, quality outcomes, data-flow and exit audits; one case appears | Tool changes fail to alter behavior when policy, incentives, and staffing remain stable | tool, culture, blueprint | EPUB-C08#sec1–#sec12 | Q3, Q13, Q26 | | C-013 | EA | AI can assist research tasks but should not replace accountable human judgment | Evaluative, time-sensitive | Task-specific evaluations, error/harm rates, human-review tests; 2023 survey appears | Validated systems outperform accountable humans on defined tasks without added harm or loss of recourse | AI, accountability | EPUB-C08#sec11 | Q26, Q27 | | C-014 | EA | Ethics, privacy, and governance must operate together, with ongoing consent | Moral and practical | Comprehension, withdrawal, deletion, incident, and remedy outcomes; cases and codes appear | Separated controls protect people equally across the data lifecycle | consent, governance | EPUB-C09#sec1–#sec13 | Q18, Q20, Q26 | | C-015 | EA | Budgets and metrics make operations governable and help secure investment | Practical and causal | Forecast accuracy, decision use, outcome linkage, gaming audits; examples appear | Added reporting increases work or gaming without improving allocation or outcomes | metric, budget | EPUB-C10#sec1–#sec11 | Q9, Q27 | | C-016 | EA | Prioritization must shape demand through explicit contracts, criteria, and tracking | Practical, organizational | Conflict, cycle, appeal, alignment, and protected-capacity outcomes; cases appear | Transparent systems produce no better decisions and worsen power imbalance | priority, contract | EPUB-C11#sec1–#sec15 | Q15, Q16, Q18 | | C-017 | EA | Research talent requires competency, inclusive hiring, paced onboarding, and growth systems | Practical, workforce | Hiring, pay, promotion, retention, access, and appeal outcomes; examples appear | Structured systems preserve or worsen bias and produce no improvement in retention or growth | competency, DEI | EPUB-C12#sec1–#sec13 | Q20, Q24 | | C-018 | EA | Recurring protected reflection is required to review and renew the whole system | Practical, role-dependent | Decision quality, incident discovery, maintenance and adaptation records; one author example appears | Teams without protected reflection adapt equally well without greater failure or burnout | reflection, review | EPUB-CODA | Q10, Q11 | ### Evidence-expectation pattern The book supplies definitions, logical arguments, 146 notes, first-person cases, selected third-party cases, trade sources, professional frameworks, and worked examples. It rarely supplies comparative cohorts, complete denominators, raw data, independent replication, participant outcome measures, or failed implementations. Evidence is strongest for **what to inspect and design**, moderate for **mechanism plausibility**, and weak for **universal effectiveness or causal magnitude**. ### Claim dependency graph ```mermaid flowchart TD C1["C-001/C-002: systems absorb changing demand"] --> C3["C-003: research operating system"] C3 --> C4["C-004: paired strategies"] C4 --> C6["C-006: strategy changes operations"] C6 --> C7["C-007: eight interdependent elements"] C7 --> C8["C-008: lifecycle planning"] C8 --> INFRA["C-009–C-013: recruitment, knowledge, support, tools, AI"] C8 --> GOV["C-014–C-016: respect, money, priorities"] C8 --> PEOPLE["C-017/C-018: talent and reflection"] INFRA --> OUT["Scalable, useful research practice"] GOV --> OUT PEOPLE --> OUT C5["C-005: portfolio-level value"] --> C15["C-015: money and metrics"] C15 --> OUT classDef weak stroke-dasharray: 5 5,stroke:#b45309; class C2,C3,C6,C15 weak; ``` **Load-bearing claims:** C-001/C-002, C-003, C-004, C-006, C-007, and C-008. The practical chapters can remain useful if one taxonomy is incomplete. The book's broad title-level argument weakens if system design does not preserve value under changing demand. C-009 through C-018 can fail independently without collapsing every other chapter. ### Claim-to-question map - **Scale and system:** C-001–C-003 → Q1–Q3, Q11, Q27 - **Strategy and value:** C-004–C-006 → Q3, Q6, Q9, Q10, Q27 - **Framework and lifecycle:** C-007–C-008 → Q4, Q5, Q12, Q13 - **Infrastructure:** C-009–C-013 → Q5–Q7, Q13–Q14, Q17–Q18, Q20, Q23, Q26 - **Governance and allocation:** C-014–C-016 → Q9, Q15–Q18, Q20, Q26–Q27 - **People and renewal:** C-017–C-018 → Q10–Q11, Q20, Q24 --- ## Prompt 5 — quantitative rigor and confidence updates ### Policy The numerical ranges below encode judgment; they are not measured probabilities. An update requires a defined claim, relevant evidence, an explicit dependence check, and a reason the evidence would be more or less likely if the claim were false. Multiple passages from Towsey's experience remain one dependent evidence stream. A cited case can be independent of Towsey but still depend on a vendor, consultant, or success-story selection process. | Label | Range | |---|---:| | Very low | 0–20% | | Low | 21–40% | | Uncertain | 41–60% | | Moderate | 61–75% | | High | 76–90% | | Very high | 91–100% | ### Priors and posteriors | Claim | Prior | Evidence considered | Impact | Posterior | Band | Dominant constraint | |---|---:|---|---|---:|---|---| | C-001 Systems, not volume, determine scale | 75% | Definition, demand examples, cross-chapter fit, operating theory | Moderate increase | 82% | High | Few comparative outcome studies | | C-004 Research and ops strategies should align | 80% | Logical what/how link, author cases, strategy literature | Small increase | 84% | High | Agreement is partly conceptual rather than causal | | C-007 All eight elements matter | 55% | Recruitment and training dependency walkthroughs | Small increase | 63% | Moderate | Completeness of the taxonomy remains untested | | C-008 Lifecycle planning supports sustainability | 75% | Build/improve/KTLO model and maintenance examples | Small increase | 80% | High | No longitudinal comparison or retirement data | | C-009 Standardized recruitment can improve operations | 60% | Ten-step workflow, failed self-service case, vendor survey, inclusion checks | Moderate increase | 70% | Moderate | No matched equity and quality outcomes | | C-010 RKM needs stewardship beyond a repository | 85% | Knowledge theory, failure analysis, Atlassian case, vault-adjacent evidence | Small increase | 88% | High | Organizational case evidence remains selected | | C-012 Tools shape culture | 70% | Tool-removal case, workflow mapping, behavior mechanism | No meaningful update | 71% | Moderate | Causality is not isolated from policy and staffing | | C-014 Respect controls must be integrated | 90% | Data-lifecycle logic, Facebook case, professional codes, consent analysis | Small increase | 92% | Very high | High confidence concerns the principle, not legal implementation | | C-015 Metrics improve governance | 65% | Budget/report examples and decision logic | No meaningful update | 65% | Moderate | Proxy gaming and missing outcome links | | C-016 Explicit priority systems reduce conflict | 75% | Power analysis, NASA/GDS examples, collaboration model | Small increase | 79% | High | Few direct conflict or appeal outcomes | | C-017 Competency and onboarding systems improve talent outcomes | 60% | HR frameworks and selected organization examples | Small increase | 64% | Moderate | Promotion, compensation, bias, and retention outcomes absent | | C-018 Protected reflection improves system renewal | 60% | One quarterly practice and plausible mechanism | Small decrease | 58% | Uncertain | Single privileged-role example | ### Calibration warnings | Warning | Severity | Application | |---|---|---| | False precision | High | Percentages are judgment aids; they do not measure truth | | Double-counted evidence | High | Repeated author cases and chapter agreement are one evidence stream | | Correlated sources | High | Practitioner networks, vendor reports, and publisher-selected cases may share incentives | | Publication bias | High | Successful or memorable cases are easier to publish than failed operations | | Survivorship bias | High | Google, Atlassian, GDS, H&M, IDEO, and mature teams dominate examples | | Base-rate neglect | Moderate | The book rarely states how often practices succeed across all attempts | | Tiny samples | High | The strategy poll has 166 self-selected respondents; several claims rest on one case | | Selection effects | High | Survey respondents and community members are already engaged with research practice | | Motivated estimates | Moderate | Author, vendors, and professional community benefit from ResearchOps legitimacy | | Unavailable denominators | High | Metrics often lack total eligible users, studies, failures, or excluded participants | | Proxy measures | High | Attendance, report views, counts, and spend can rise without better research or decisions | | P-values as effect sizes | Not applicable | The book does not present inferential tests | | Statistical versus practical significance | Low | Few formal statistics appear; practical meaning remains the larger issue | ### Qualitative fallback - **Strongly increases confidence:** Independent, multi-context outcome evidence with defined denominators and harms. - **Moderately increases confidence:** A transparent case with baseline, decision trace, costs, limits, and participant outcomes. - **No meaningful update:** Another unmeasured author anecdote or repeated chapter assertion. - **Moderately decreases confidence:** A credible counterexample showing hidden cost, exclusion, or no change. - **Strongly decreases confidence:** Comparative evidence that the intervention causes worse research, participant harm, or durable power imbalance. --- ## Prompt 6 — analytical scaffold ### Chapter analysis slots The `Response` column is a **system response**, not a reader reaction. | Chapter | Purpose and thesis | Claims and terms | Evidence and example | Counterexample, assumptions, unanswered question | Spine, anchors, and response | |---|---|---|---|---|---| | 1 | Define scale; systems must absorb demand without losing value | C-001–C-003; scale, operating system, culture | Gift-volume, growing-team, transit, and recruitment cases | No failed transit-like system; assumes efficiency and sustainability; what thresholds count? | Foundation; EPUB-C01#sec3–#sec12; 🔶 useful reframe, absolutes need repair | | 2 | Strategy selects purpose and aligns ResearchOps | C-004–C-005; research strategy, value, trust | 166-person poll, layoff report, strategy literature | Self-selected poll; executive perception may displace truth; whose value controls? | Purpose; EPUB-C02#sec1–#sec15; 🔶 strongest when refusal is explicit | | 3 | Strategy becomes operating design | C-006; operating model, service pairs | H&M/IDEO, GDS lab, author cases | Binary pairs hide mixtures; H&M attribution incomplete; what alternatives explain results? | Bridge; EPUB-C03#sec1–#sec15; 🔶 practical design vocabulary | | 4 | Eight elements and three phases scope work | C-007–C-008; elements, maturity, KTLO | Recruitment-panel and training walkthroughs | No counterexample in chapter; grid hides recursion and retirement; is taxonomy complete? | Scaffold; EPUB-C04#sec1–#sec9; 🔶 valuable checklist, not universal ontology | | 5 | Standardize recruitment as an inclusive participant journey | C-009; taxonomy, targeting, screening, consent | Failed self-service case, route comparison, 2023 vendor survey | Central categories can exclude; local legal rules vary; who governs cohorts? | Infrastructure; EPUB-C05#sec1–#sec22; 🔶 ethically strong, needs equity outcomes | | 6 | RKM joins communities and governed knowledge services | C-010; tacit/explicit, taxonomy, repository, RKM | Knowledge scholarship and Atlassian library case | False binary between community and service; arithmetic/source defects; who defines official knowledge? | Infrastructure; EPUB-C06#sec1–#sec12; 🔶 one of the strongest chapters | | 7 | Support needs journeys, channels, and capacity controls | C-011; logistical/craft support, SLA, PWDR | Channel models and author experience | Training gates can block urgent access; private cases need private routes; how is burden measured? | Infrastructure; EPUB-C07#sec1–#sec22; 🔶 transferable with safety exceptions | | 8 | Tools encode culture and require lifecycle governance | C-012–C-013; tooling blueprint, ResTech, AI | Tool-removal case, 274-tool list, 1,093-person vendor survey | Tool effect confounded; AI snapshot dated; who represents participants in selection? | Infrastructure; EPUB-C08#sec1–#sec12; 🔶 strong procurement model, weak AI policy | | 9 | Ethics, privacy, and data governance must operate together | C-014; consent, PII, DSAR, RTBF | Facebook study, author harm cases, codes and laws | Controls can become theater; no outcome evaluation; how are harms remedied? | Governance; EPUB-C09#sec1–#sec14; 🔶 moral center of the book | | 10 | Budgets and metrics fund and reveal operations | C-015; cost center, OpEx/CapEx, dashboard | Budget practice, vendor case, sample reports | Metrics can game work; no equity budget; what action changed because of a number? | Feedback; EPUB-C10#sec1–#sec11; 🔶 adopt only with quality guardrails | | 11 | Prioritization manages demand, contracts, and power | C-016; engagement, collaboration model, backlog, PgMT | GDS, NASA, author cases | Neutral-looking criteria may preserve incumbent power; who can appeal? | Allocation; EPUB-C11#sec1–#sec15; 🔶 explicit power analysis is a strength | | 12 | Talent systems support capability, inclusion, onboarding, and growth | C-017; competency, pipeline, DEI, onboarding | HR sources and IDEO/Figma/GDS examples | Referrals and scoring can preserve bias; no pay/promotion audits; whose work counts? | People; EPUB-C12#sec1–#sec13; 🔶 useful scaffold with major labor gaps | | Coda | Protected review time renews the system | C-018; reflection cadence | Author's quarterly “big picture” week | No counterexample; assumes schedule control; how can frontline roles participate? | Renewal; EPUB-CODA; 🔶 adapt to shared quiet time and rotating coverage | ### Argument spine 1. **Starting assumptions:** Research is variable, knowledge-producing, and socially situated; demand exceeds available resources. 2. **Foundational definitions:** Scale is system response under demand; ResearchOps designs the surrounding system. 3. **Major premises:** Purpose must precede infrastructure; research and operations strategies must align; operating choices shape culture, access, cost, and quality. 4. **Intermediate conclusion:** The eight elements and maturity phases expose dependencies across people, services, data, tools, rules, and maintenance. 5. **Central conclusion:** Research scales responsibly only when its operating system preserves useful judgment, trust, participant respect, and adaptability. 6. **Practical recommendations:** Define priorities and refusal; map journeys and data; select operating models; assign owners; measure outcomes and harms; fund maintenance; review and retire systems. **Weakest link:** The book moves from coherent system logic and selected cases to broad claims of effectiveness without comparative outcome evidence. Repair it with transparent local pilots, failure data, participant and worker outcomes, and cross-context comparisons. The repair keeps the handbook useful but reduces its rhetorical certainty. ### Evidence ledger | Evidence | Claims | Type and source | Independence | Grade | Limitations | Anchors | |---|---|---|---|---|---|---| | E-001 | C-001–C-003 | Author demand and recruitment cases | No | C | Illustrative, selected, no baseline | EPUB-C01 | | E-002 | C-004 | 2024 LinkedIn poll, 166 respondents | Partial | D | Self-selected, no population frame | EPUB-C02#sec1 | | E-003 | C-004–C-006 | Strategy and operations literature | Yes, conceptually | B | General business transfer | EPUB-C02 notes; EPUB-C03 notes | | E-004 | C-006 | H&M/IDEO pilot, 22% stock reduction and 34% sales increase | Partial | C | Vendor case; causal path and emissions outcome unclear | EPUB-C03#ch3 | | E-005 | C-007–C-008 | Recruitment-panel and training walkthroughs | No | C | Worked examples, not outcome tests | EPUB-C04 | | E-006 | C-009 | Author recruitment failure and route comparison | No | C | Strong mechanism, one practitioner stream | EPUB-C05#ch5–#sec10 | | E-007 | C-009 | User Interviews 2023 recruitment survey | Partial | C | Vendor audience and methods limit transfer | EPUB-C05 note 5 | | E-008 | C-010 | Knowledge-management scholarship | Yes | B | Theory does not prove this exact operating model | EPUB-C06 notes | | E-009 | C-010 | Atlassian library case | No | C | No comparative costs or decision outcomes | EPUB-C06#sec1 | | E-010 | C-011–C-012 | Support and tool cases | No | C | Confounded by staffing, policy, and leadership | EPUB-C07; EPUB-C08 | | E-011 | C-013 | User Interviews 2023 AI survey | Partial | C | Vendor sample and dated capability snapshot | EPUB-C08#sec11 | | E-012 | C-014 | Facebook emotional-contagion study and ethics criticism | Yes | B | Strong counterexample; commercial research transfer still contextual | EPUB-C09#ch9–#sec4 | | E-013 | C-014 | Ethics, privacy, and legal codes cited by book | Yes | B | Jurisdiction and revision drift | EPUB-C09 notes | | E-014 | C-015 | Author budget/report practice | No | C | No measured governance outcome | EPUB-C10 | | E-015 | C-016 | GDS and NASA prioritization examples | Yes, selected | B | Different institutions and outcome definitions | EPUB-C11 notes | | E-016 | C-017 | HBR/Gallup onboarding summary and company examples | Partial | C | Indirect source chain and selection effects | EPUB-C12#sec3–#sec9 | | E-017 | C-018 | Quarterly reflection practice | No | D | One high-autonomy role | EPUB-CODA | Six ledger entries have meaningful independence from Towsey: E-003, E-008, E-012, E-013, E-015, and part of E-016. E-012 tells against any claim that legal permission or platform consent alone settles ethics. No independent comparative source directly tests the book's complete operating system. ### Reading plan and Prompt 7 entry criteria - **Order completed:** Front matter → Ch. 1–4 in sequence → Ch. 5–12 → coda → notes and figures. - **Highest attention:** Ch. 1–6, 8–11; these carry the argument, infrastructure, risk, and allocation model. - **Lower attention:** Praise, promotional back matter, index entries, repeated “In a Nutshell” sections after full chapter reading. - **External checks:** Publication identity, ResearchOps lineage, current professional ethics, AI regulation, and accessibility context. - **Comparison terms:** ResearchOps pillars/elements, sociotechnical systems, knowledge service/repository, consent, program management, impact. - **Synthesis stops used:** End of Ch. 4, Ch. 8, Ch. 11, and coda. Prompt 7 began only after all chapters were read; C-001–C-018, E-001–E-017, source labels, falsifiers, confidence rules, evidence dependence, and corrections were present. Criteria satisfied. --- ## Prompt 7 — x-ray analysis and fair criticism ### Strongest reconstruction **Author says:** Research knowledge cannot be mass-manufactured, but the conditions around it can be designed. When demand and variation grow, a research practice needs explicit purpose, paired research and operations strategies, context-fit operating models, and planning across eight interdependent elements and the full lifecycle. Recruitment brings people into the system; knowledge management preserves what is learned; support keeps the system usable; tools encode behavior and data movement; respect protects participants; money and metrics sustain and observe work; prioritization allocates scarce capacity; talent systems sustain the people doing the work. Culture, trust, feedback, and protected reflection join these elements. **We infer:** This reconstruction is more linear than the book's many frameworks and stories, but it remains faithful to their dependency structure. ### Claim evaluation | Claim | Support | Grade | Strongest objection and counterexample | Current conclusion | |---|---|---:|---|---| | C-001 | Definitions, capacity examples, chapter coherence | B | Some research networks expand through informal norms without formal ops | Strong reframe; scope to variable knowledge work | | C-002 | Logical definition and diagrams | C | A system may absorb demand while externalizing cost or harm | Require explicit thresholds for quality, access, cost, and harm | | C-003 | Transit/city models and cases | C | “Complete” is unreachable; culture can resist coherent systems | Use as design criteria, not proof of completeness | | C-004 | Strategy logic, literature, cases | B | Two strategies can split decisions or add document theater | Keep one joined decision system with what/how fields | | C-005 | Portfolio logic and cost-center framing | C | Some experiments permit credible study-level attribution | Prefer portfolio judgment but preserve direct causal evidence when available | | C-006 | H&M and author cases | C | Goals can succeed through market or leadership changes unrelated to ops | Changed operations are usually necessary, not sufficient | | C-007 | Dependency walkthroughs | C | Accessibility, security, procurement, communications, and decommissioning may deserve separate lenses | Eight elements are a useful minimum map, not a closed set | | C-008 | Lifecycle model and maintenance logic | B | Systems can fail despite planning; maturity can reverse | Adopt with incident and retirement stages | | C-009 | Detailed workflow, cases, survey | B | Taxonomies and standards can exclude people outside known cohorts | Adopt with community input, exceptions, and equity audits | | C-010 | Scholarship, failure modes, Atlassian case | B | Small teams may gain enough value from a light repository | Stewardship is required; its form should match risk and scale | | C-011 | Service models and capacity mechanisms | B | Open peer networks can outperform formal channels for some questions | Design a mixed service with private safety routes | | C-012 | Tool case and plausible mechanism | C | Incentives and policy may dominate tool effects | Treat tools as one cultural control, not the cause | | C-013 | 2023 survey and accountability principle | C | Some bounded tasks may become safer with automated checks than with unreviewed humans | Keep task-specific evaluation and named human accountability; recheck capability | | C-014 | Data logic, harm cases, current ethics context | B | Integrated programs can still become compliance theater | Adopt principle; test comprehension, remedy, and outcomes | | C-015 | Budget practice and reporting examples | C | Dashboards can create gaming and surveillance | Use measures only with decision links and guardrails | | C-016 | Power analysis, GDS/NASA, models | B | Formal criteria can legalize incumbent preferences | Co-design, publish decisions, protect appeals and exploration | | C-017 | HR frameworks and cases | C | Competency scoring and referrals can reproduce bias | Pilot with pay, promotion, access, and appeal audits | | C-018 | Plausible mechanism and one author example | D | Frontline workers may lack schedule autonomy; shared review rituals may work better | Adapt to institutionally protected reflection; evidence remains thin | ### Grade distribution - **A:** 0 - **B:** 8 - **C:** 9 - **D:** 1 - **F:** 0 - **N/A:** 0 The pattern matters: the book grades best where a practical mechanism is supported by professional theory, detailed workflow, and a relevant case. It grades worse where the author makes broad causal, completeness, metric, AI, or transfer claims. No claim earns A because none combines direct outcome evidence, clear denominators, independent support, and strong cross-context comparison. ### Biases, fallacies, and reasoning risks | Issue | Evidence | Strength | Judgment | |---|---|---:|---| | Selection and survivorship bias | Large, recognizable, successful organizations dominate examples | Material | Genuine evidence-selection risk, not proof the practices fail | | Hasty generalization | 166-person LinkedIn poll informs a profession-wide strategy gap | Moderate | The poll is suggestive, not representative | | Correlation-causation | H&M results and tool-removal culture case | Moderate | Narrative attribution exceeds the data shown | | Equivocation | Scale, value, trust, culture, impact | Material | Meanings change enough to affect measurement and falsification | | Appeal to metaphor | City, transit, ecosystem, periodic table | Low | Rhetorical aid, not a formal fallacy unless treated as evidence | | Authority bias | Google, Atlassian, IDEO, GDS, NASA | Moderate | Named institutions raise credibility without supplying full methods | | Base-rate neglect | Success cases lack all-attempt denominators | Material | Effectiveness prevalence remains unknown | | Proxy substitution | Views, attendance, counts, spending, visibility | Material | The book warns about interpretation but does not supply outcome validation | | False precision | 1% compounding, chapter arithmetic, gravitational constant | Local | Three factual errors do not defeat the framework but require correction | | Unfalsifiable framing | Culture can explain success or failure after the fact | Moderate | Repair by naming observable behaviors and rival explanations | ### Counter-observations in the author's favor - Towsey repeatedly says the frameworks are adaptable and not rigid (`EPUB-INTRO`). - She distinguishes legal compliance from ethics and warns against mechanistic care (`EPUB-C09`). - She reports a failed self-service recruitment system rather than presenting only successes (`EPUB-C05`). - She notes that metrics can be missing, inaccurate, incomplete, and slow to become useful (`EPUB-C10`). - She argues against tool-first operations and against replacing participants with synthetic users (`EPUB-C05`, `EPUB-C08`). - She concedes that one person cannot master or own every ResearchOps specialty (`EPUB-C04`). - She frames operating models as contextual choices, not one best structure (`EPUB-C03`). ### Repairs | Weak claim | Repair | Faithful? | Cost | |---|---|---|---| | All scalable systems are efficient and sustainable | A responsibly scalable system stays within declared quality, access, cost, resilience, and harm limits over a tested demand range | Yes | Loses memorable absolutes | | Every initiative involves all eight elements | Use the eight as a minimum dependency scan; add local cross-cutting controls and record non-applicable cells | Yes | Gives up taxonomy completeness | | Tools shape culture | Tools contribute to culture by changing defaults and constraints, alongside incentives, authority, training, and staffing | Yes | Reduces causal force | | Metrics create value and investment | Measures can improve allocation when tied to outcomes, decisions, denominators, and anti-gaming checks | Yes | Adds work and uncertainty | | AI cannot replace researchers | Current systems must be evaluated task by task; no system replaces accountable judgment, participant relationships, and recourse as one bundle | Yes | Requires frequent reassessment | | Reflection is individually created | Leadership must create equitable, institutionally protected review time and coverage | Yes | Changes a personal-discipline story into an organizational obligation | No major claim is beyond repair. The least repairable parts are numerical errors and claims of broad transfer from thin evidence; they should be corrected or narrowed, not defended. ### Overall critique **We conclude:** This is a strong operating handbook and a moderate evidence source. Its frameworks make hidden dependencies visible and encourage lifecycle, ethics, support, and maintenance thinking. It is not a validated maturity model, causal theory of organizational value, current AI policy, or jurisdiction-specific compliance guide. --- ## Prompt 8 — context, systems, and intellectual lineage ### Context available to the author - **Historical:** ResearchOps gained a shared label and community in the late 2010s, while related work existed earlier inside large organizations. - **Institutional:** The book draws heavily from corporate UX research, ResearchOps Community work, Atlassian, GDS, Google, and professional networks. - **Economic:** It responds to 2020–24 pandemic disruption, research-team growth, layoffs, budget scrutiny, and pressure to prove value. - **Technological:** It was written during rapid expansion of research platforms and the first public wave of generative AI tools. - **Disciplinary:** It joins operations management, service and systems design, knowledge management, research ethics, HR, finance, and program management. - **Cultural:** It assumes organizations value human-centered research but struggle to coordinate, fund, govern, and reuse it. ### Intellectual lineage ```mermaid flowchart LR OM[Operations management] --> OPS[DevOps / DesignOps / ResearchOps] STS[Sociotechnical systems] --> SD[Systems and service design] SD --> ROS[Research operating system] KM[Knowledge management] --> RKM[Research knowledge management] ETH[Research ethics and data protection] --> RESPECT[Respect-in-research controls] HR[People operations] --> TALENT[Research talent systems] FIN[Finance and program management] --> GOV[Budget, metrics, priorities] OPS --> ROS RKM --> ROS RESPECT --> ROS TALENT --> ROS GOV --> ROS ``` The current [ResearchOps Community](https://researchops.community/about/) still presents eight pillars—environment, scope, people, organizational context, recruitment/admin, data/knowledge management, governance, and tools/infrastructure—and describes ResearchOps as an emergent profession. Towsey's eight *elements* overlap but regroup the field around operational delivery. The connection is documented, not an anachronistic inference. ### Systems and incentives - Research leaders seek capacity, influence, quality, and protection during budget pressure. - Operations practitioners gain professional identity and scope when invisible work becomes a named discipline. - Vendors benefit when organizations see recruitment, repositories, support, and tooling as durable infrastructure. - Executives may reward visible efficiency and strategic alignment more readily than negative findings, slow learning, or harms prevented. - Participants provide the data and experience on which the system depends but often receive the least decision authority. - Centralization can improve consistency and safety while concentrating classification, access, funding, and priority power. ### Later developments added after publication - The [2025 ICC/ESOMAR International Code](https://standards.esomar.org/assets/documents/icc-esomar-code-2025.pdf) updates professional research guidance around AI, synthetic data, data minimization, duty of care, transparency, responsibility, and human oversight. It strengthens the need to revise Ch. 8–9 controls beyond the book's 2023 AI snapshot. - The [EU AI Act implementation page](https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai) shows that general-purpose AI obligations became applicable in August 2025 and broader rules and transparency duties applied in August 2026, with later dates for some high-risk systems. This does not make every research tool high risk, but it makes jurisdiction and use-case review necessary. - [WCAG 2.2](https://www.w3.org/TR/WCAG22/) remained the current W3C recommendation and became ISO/IEC 40500:2025 after the book's publication. Accessibility needs a test plan, not a vendor checkbox. - ResearchOps practice continues to change with layoffs, distributed research, AI-assisted synthesis, privacy regulation, and vendor consolidation. Named tools and labor-market statements require review. ### Perspectives shaping and missing from the record **Present:** Research leaders, ResearchOps practitioners, large-company operators, UX/product disciplines, finance, HR, legal/privacy partners, selected participants, and knowledge-management thinkers. **Underrepresented or missing:** Participants as co-governors; disabled and neurodivergent researchers in sustained first-person depth; Indigenous data sovereignty; labor organizers; small nonprofits and volunteer communities; Global South institutions outside large technology networks; adversarial findings that leaders reject; failed operations teams; security and records-management specialists; worker privacy in dashboards and competency systems. ### Historical judgment The book captures ResearchOps at a professionalizing moment: the field is building shared language, jurisdiction, methods, and career paths while responding to cost pressure and AI disruption. That context explains both its strength and its risk. Naming hidden work makes it governable; professionalizing it can also expand scope, centralize power, and turn local practices into a marketable universal model. --- ## Prompt 9 — DEIB, multicultural, accessibility, and reading-level audit ### Inclusion strengths - Recruitment guidance calls for disability access, plain language, income diversity, device choice, agency scrutiny, and deliberate panel composition (`EPUB-C05#sec21`). - Ch. 9 treats consent as ongoing and calls for accessible and translated information, withdrawal, deletion, support, anonymity, and care for participants and researchers. - Tool criteria include WCAG, permissions, privacy/security, data ownership, APIs for deletion, and vendor labor practices (`EPUB-C08#sec8`). - Talent guidance names inclusive job language, structured hiring, accessibility of tools and onboarding, bias risks, and targeted outreach (`EPUB-C12#sec8`). - The author uses “research practice” to include people outside formal teams and recognizes varied professional backgrounds. - The strongest structural mechanism is the repeated instruction to build inclusion, consent, privacy, and support into workflows rather than rely on stated values. ### Exclusion and omission risks - Most cases assume corporate product organizations with dedicated HR, legal, finance, security, procurement, and tooling capacity. - The protected-characteristic list reflects a UK frame and does not replace local law, intersectional analysis, or community-defined categories. - Participants receive safeguards but little formal power over taxonomies, data reuse, knowledge products, priorities, or vendor selection. - Disability often appears as an accommodation, specialist cohort, or vendor criterion rather than a source of operating authority. - Neurodiversity receives little sustained treatment in research timing, sensory load, communication, synthesis, or employment systems. - Inclusion goals lack consistent outcome measures for invitation, participation, completion, compensation, harm, promotion, retention, or appeals. - Referral hiring, public speaking, and team branding can favor people with existing networks, schedule freedom, English fluency, and public visibility. - Research dashboards and study trackers can expose worker or participant data; privacy guidance focuses more on participants than staff surveillance. - “Go it alone” DEI advice can put sensitive institutional work on an underqualified individual without power or protection. ### Multicultural transferability | Context | Transfers well | Requires adaptation | Main risk | |---|---|---|---| | Small or volunteer team | Purpose, dependency scan, maintenance, support routes | Combine roles and reduce artifacts | Enterprise process overwhelms capacity | | Open-source community | Explicit intake, knowledge stewardship, transparent priorities | Consent, public/private boundaries, unpaid labor, shared governance | Contributors treated as free participants or staff | | Public sector | System maps, accessibility, records, program management | Procurement, public law, political accountability, records retention | Business-value framing displaces public value | | Academia | Ethics, knowledge lifecycle, participant care | IRB/REC rules, publication incentives, disciplinary methods | Corporate speed and value assumptions | | Global-majority organization | Context-fit design and local partnerships | Language, infrastructure, compensation, data sovereignty, local law | Anglo-American categories and vendor dependency | | Agency/consultancy | Collaboration models, budgets, recruitment, knowledge handoff | Client ownership, deletion, reuse, power, short engagements | Conflicting duties to client and participant | | Disability-led research | Accessible standard paths and expert support | Shared decision power, compensating expertise, varied communication | Treating access as an exception service | | High-risk or regulated field | Ethics/privacy/governance integration | Specialist review, incident response, legal duties, audit evidence | Handbook used as a compliance substitute | ### Accessibility audit of the supplied EPUB | Area | Finding | |---|---| | Structure | Valid XHTML, declared reading order, navigation, headings, page markers, and table of contents | | Language | Publication language declared `en-US`; package admits language shifts were not checked | | Images | 45 content images have nonempty alt text; 9 decorative images have empty alt text and presentation role | | Complex diagrams | Alternatives summarize nodes and relationships but do not always provide full data or reading order | | Color | Package says image contrast was not checked; no independent contrast test performed | | Reflow | EPUB declares display transformability and textual access modes | | Conformance claim | EPUB Accessibility 1.1 / WCAG 2.2 Level A; statement says generally produced toward AA, not fully checked | | Reference navigation | Original XHTML footnote targets exist; Pandoc's merged conversion generated reference warnings, so converted Markdown is not an accessibility proxy | | Assistive technology | Not tested with VoiceOver or another screen reader; no claim of practical conformance is made | ### Reading demand - **Syntactic difficulty:** Moderate. Sentences are conversational, examples are concrete, and summaries recur. - **Reference density:** Moderate to high. Operations, finance, HR, knowledge-management, privacy, service-design, and research acronyms accumulate. - **Conceptual load:** High in Ch. 1–4 and Ch. 6 because several metaphors and models overlap. - **Estimated level:** Grade 10–12 for prose; practitioner or undergraduate familiarity improves comprehension. - **Access barriers:** Book price, time, 90,000-plus words, business jargon, image-heavy frameworks, English, named enterprise tools, and assumed access to specialist partners. ### Supplementary perspectives - Participant-led and community-based participatory research. - Indigenous data sovereignty and CARE principles alongside FAIR data practices. - Disability-led research and inclusive research methods. - Labor, worker privacy, compensation equity, and organizational surveillance. - Records management, information security, and incident response. - Small-team, nonprofit, open-source, public-sector, and Global South ResearchOps cases. - Comparative evaluations of repositories, recruitment services, support models, and competency systems. --- ## Prompt 10 — notes pipeline and durable knowledge ### Fleeting notes - Test whether the Planning Matrix needs a fourth phase: retire, migrate, or decommission. - Compare Towsey's elements with the ResearchOps Community pillars and identify coverage differences. - Ask whether “culture” can be replaced by observable behaviors, incentives, and decisions in each experiment. - Find a failed research-library case with enough detail to test the stewardship argument. - Separate participant inclusion from participant governance. - Add a private escalation path anywhere support is moved out of direct messages. - Treat taxonomies as versioned decisions with owners and appeals. - Check whether a study backlog has a stale-item rule before connecting it to metrics. - Revisit AI sections only with task-specific, current evidence. - Never treat framework completion as proof of research quality. ### Literature notes | Suggested filename | Source claim | Concise note | Attribution | |---|---|---|---| | `Towsey — Scale the Research System, Not Research Judgment.md` | C-001–C-003 | Research remains variable; routes, support, tools, rules, and maintenance absorb changing demand | Author argument, EPUB-C01 | | `Towsey — Strategy Before ResearchOps Infrastructure.md` | C-004–C-006 | What the practice chooses must govern how operations spend, standardize, and refuse | Author argument, EPUB-C02–C03 | | `Towsey — ResearchOps Planning Matrix.md` | C-007–C-008 | Scan every initiative across operational elements and lifecycle phases | Author framework, EPUB-C04 | | `Towsey — Recruitment Is a Participant Lifecycle.md` | C-009, C-014 | Recruitment runs from brief through consent, deletion, and continuing support | Author workflow, EPUB-C05 and C09 | | `Towsey — Communities and Knowledge Services.md` | C-010 | Social knowledge creation and governed durable access need different but connected systems | Author synthesis, EPUB-C06 | | `Towsey — Research Support as a Service System.md` | C-011 | Support channels need audience, risk, capacity, maintenance, and escalation design | Author framework, EPUB-C07 | | `Towsey — Tooling Blueprint and Data Movement.md` | C-012–C-013 | Tool choice encodes workflow, permissions, data movement, support, cost, and exit | Author framework with analysis limits, EPUB-C08 | | `Towsey — Respect in Research.md` | C-014 | Ethics, privacy, governance, and consent must operate as one lifecycle | Author argument, EPUB-C09 | | `Towsey — Metrics and Priority Power.md` | C-015–C-016 | Visibility and prioritization allocate resources and authority; guardrails are required | Author argument plus analysis, EPUB-C10–C11 | | `Towsey — Talent and Reflection as System Maintenance.md` | C-017–C-018 | Hiring, onboarding, growth, and review time sustain the human system | Author argument, EPUB-C12 and coda | ### Permanent-note candidates These are candidates, not files created during this one-file run. #### A participant taxonomy is a governed hypothesis - **Suggested filename:** `A Participant Taxonomy Is a Governed Hypothesis.md` - **Claim:** A reusable participant taxonomy improves routing only when the organization treats its categories as revisable claims about who matters, not as a fixed description of reality. - **Confidence:** Moderate. - **Scope and conditions:** Applies to recurring research recruitment with centrally named cohorts. Requires purpose, version owner, missing-population log, intersectional audit, exception route, and community review. Does not reject all segmentation. - **Evidence:** EPUB-C05#sec1–#sec21; C-009; E-006–E-007. - **Uncertainty:** The EPUB gives no comparative equity outcome. - **Links:** [[3. Resources/Permanent Notes/Retrieval Quality Is Set by the Unit of Knowledge Not the Retrieval Algorithm]]; [[3. Resources/Permanent Notes/Sensitive Research Needs Separate Categories for Testimony Allegation and Finding]]. #### A tool purchase is a data, labor, and exit policy - **Suggested filename:** `A Tool Purchase Is a Data Labor and Exit Policy.md` - **Claim:** Selecting a research tool commits an organization to permissions, data movement, participant terms, support work, vendor labor, maintenance, and an eventual exit path—not just software features. - **Confidence:** High for the dependency claim; moderate for outcome effects. - **Scope and conditions:** Applies when a tool stores research data or coordinates multi-person work. Weaker for disposable, local, non-sensitive tools. - **Evidence:** EPUB-C08#sec3–#sec10; C-012; E-010. - **Uncertainty:** The book does not compare lifecycle-governed and feature-first purchases. - **Links:** [[3. Resources/Permanent Notes/An Organization Cannot Expose Data to a New System It Cannot Locate Internally]]; [[3. Resources/Permanent Notes/Adopting an Ecosystem Component Inherits Its Maintenance Risk]]. #### Operational visibility needs quality and harm guardrails - **Suggested filename:** `Operational Visibility Needs Quality and Harm Guardrails.md` - **Claim:** An operations metric becomes decision-worthy only when the same view shows what quality, access, labor, or participant harm could worsen while the number improves. - **Confidence:** High as a decision rule; empirical magnitude unknown. - **Scope and conditions:** Applies to dashboards used for funding, staffing, prioritization, or performance. It does not require every descriptive count to predict an outcome. - **Evidence:** EPUB-C10#sec7–#sec10; C-015; E-014. - **Uncertainty:** The book does not test which guardrail set prevents gaming. - **Links:** [[3. Resources/Permanent Notes/A Metric Is Vanity Only Against Your Own Outcome Data]]; [[3. Resources/Permanent Notes/A System Can Pass Every Tracked Metric and Still Fail Its User]]. #### Support gates must not block safety access - **Suggested filename:** `Support Gates Must Not Block Safety Access.md` - **Claim:** Training and channel gates can protect expert capacity, but baseline help for ethics, accessibility, privacy, and participant safety must remain available regardless of completion status. - **Confidence:** Moderate-high. - **Scope and conditions:** Applies when a support system restricts one-to-one help or requires prior training. Extended coaching may still be gated. - **Evidence:** EPUB-C07#sec3–#sec21; C-011; E-010. - **Uncertainty:** The source provides no tested triage model. - **Links:** [[3. Resources/Permanent Notes/Training as a System Compounds Where a Session Decays]]; [[3. Resources/Permanent Notes/A Safety-Test Precondition Changes What an Incident Is Evidence For]]. #### A knowledge service needs lifecycle states, not one repository status - **Suggested filename:** `A Knowledge Service Needs Lifecycle States Not One Repository Status.md` - **Claim:** Research knowledge becomes trustworthy when each item exposes provenance, sensitivity, review state, owner, retention, supersession, and decision use. - **Confidence:** High for governance need; moderate for the proposed field set. - **Scope and conditions:** Strongest for sensitive, redundant, versioned corpora. A small low-risk collection can use a lighter schema. - **Evidence:** EPUB-C06#sec1–#sec11; C-010; E-008–E-009. - **Uncertainty:** The lifecycle extension is analysis, not Towsey's stated model. - **Links:** [[3. Resources/Permanent Notes/A Stale Backlog Creates Shadow Systems]]; [[3. Resources/Permanent Notes/Retrieval Quality Is Set by the Unit of Knowledge Not the Retrieval Algorithm]]. #### Strategy and accessibility meet in the standard path - **Suggested filename:** `Strategy and Accessibility Meet in the Standard Path.md` - **Claim:** Accessibility is strategic only when the default research journey works for disabled people; routing access needs into exceptional full service keeps exclusion inside the standard system. - **Confidence:** Moderate-high. - **Scope and conditions:** Applies to recurring recruitment, tooling, documentation, and support. Some specialist help can remain necessary. - **Evidence:** EPUB-C03#sec5–#sec13; EPUB-C05#sec21; EPUB-C07#sec8; EPUB-C12#sec8. - **Uncertainty:** The book does not report disability-led outcome testing. - **Links:** [[3. Resources/MOCs/Product & Startup Instincts MOC]]; [[3. Resources/MOCs/Book Review Synthesis MOC]]. ### Structure and connection decisions - **Book map:** This review is the source record and book-level structure note. - **Theme maps:** Strategy and refusal; research infrastructure; participant respect; knowledge governance; operational metrics; talent and labor. - **Primary MOC destinations:** [[3. Resources/MOCs/Book Review Synthesis MOC]], [[3. Resources/MOCs/Product & Startup Instincts MOC]], and [[3. Resources/MOCs/Personal Knowledge Management MOC]]. - **Related disciplines:** Product discovery, DevRel, technical communication, service design, information science, privacy, accessibility, HR, finance, and program management. - **Unresolved-question map:** Open Questions Q1–Q27 above. - **Reciprocal links:** Suggested above but not added. The one-file contract prevents creating permanent notes or editing MOCs without separate authorization. ### Attribution audit - Short quotations are marked as system-selected and carry EPUB anchors. - Claim inventory rows labeled EA summarize Towsey's claims. - E-003, E-004, E-007, E-008, E-011–E-013, E-015, and E-016 contain third-party material reported by the book; they are not independent verifications unless this review says they were checked. - Corrections to arithmetic, citation titles, and units are review findings, not author claims. - Current 2025–26 context is external analysis added after publication. - Every permanent-note candidate is system analysis and must not be quoted as Towsey. **Do not attribute to the author:** the four-cluster map; the posterior estimates; the claim grades; the lifecycle-state extension to RKM; the safety exception to training gates; the metric guardrail rule; the participant-taxonomy governance rule; the tool-as-policy formulation; the standard-path accessibility claim; the EU AI Act comparison; any vault connection or action experiment. --- ## Prompt 11 — teaching pack ### Audience and prerequisites **Audience:** Research leaders, researchers, ResearchOps and DesignOps practitioners, product managers, DevRel and technical-communication leads, knowledge managers, and graduate students studying organizational research. **Prerequisites:** Basic familiarity with user research and organizational teams. Learners do not need operations, finance, privacy, or systems-design expertise. Facilitators should define acronyms and distinguish professional guidance from empirical evidence. ### Learning objectives Learners will be able to: 1. Distinguish scaling research volume from scaling the surrounding system. 2. Connect research priorities to operations choices and refusal rules. 3. Map an initiative across operational elements and lifecycle states. 4. Identify participant, knowledge, support, data, labor, and accessibility risks. 5. Separate author experience, third-party evidence, and analysis. 6. Design one reversible local experiment with outcome and stop rules. ### Key vocabulary Scale; research operating system; research strategy; ResearchOps strategy; operating model; operational maturity; KTLO; participant taxonomy; RKM; knowledge service; logistical support; craft support; tooling blueprint; informed consent; data governance; operating metric; collaboration model; competency framework. ### Four-session sequence #### Session 1 — Purpose before infrastructure - **Book argues:** Systems, not research judgment, absorb changing demand. - **Evidence supports:** A coherent mechanism and useful definitions; broad effectiveness remains unproven. - **Activity:** Map one research service's users, demand changes, quality thresholds, and failure modes. - **Misconception check:** “Scale” does not mean “serve more people at any cost.” #### Session 2 — The eight-element lifecycle - **Book argues:** Every initiative touches eight elements and three maturity phases. - **Evidence supports:** Dependency discovery; not taxonomy completeness. - **Activity:** Apply the Planning Matrix to a research repository or participant panel, then add incident and retirement cells. - **Misconception check:** Filling every cell does not establish success. #### Session 3 — Respect, power, and knowledge - **Book argues:** Consent, ethics, privacy, governance, recruitment, and RKM are operational systems. - **Evidence supports:** Strong risk logic and useful controls; outcome evidence is incomplete. - **Activity:** Trace one participant datum from collection through use, storage, access, reuse, and deletion. - **Misconception check:** Legal permission is not the same as ethical adequacy. #### Session 4 — Metrics, priorities, and people - **Book argues:** Money, measures, collaboration contracts, competency, and reflection sustain the system. - **Evidence supports:** Decision visibility and role clarity can help; proxy gaming and bias remain live risks. - **Activity:** Redesign one dashboard metric with denominator, decision, quality guardrail, owner, and expiry. - **Misconception check:** Visibility is not impact, and a competency score is not a person. ### Discussion questions 1. Which part of research should never be standardized? 2. Who gains and loses power when operations becomes centralized? 3. Can a system be efficient but not responsibly scalable? 4. What would prove that culture changed rather than the reporting language? 5. Who should own participant and knowledge taxonomies? 6. When should craft support remain private? 7. What should a research metric allow a team to stop doing? 8. How can protected reflection include roles without schedule autonomy? ### Case studies and activities - **Failed self-service recruitment:** Diagnose why tools, training, and privacy approval did not create a navigable service (`EPUB-C05`). - **Research library:** Design lifecycle states and a retrieval test for the Atlassian case (`EPUB-C06`). - **Facebook emotional-contagion study:** Separate platform permission, legal review, informed consent, ethics, and public trust (`EPUB-C09`). - **Study tracker:** Decide which fields support planning and which create worker surveillance (`EPUB-C11`). - **Tool-selection council:** Assign decision rights to researchers, participants, privacy, security, accessibility, finance, and support (`EPUB-C08`). ### Reflection prompts - Which invisible maintenance task currently carries the most risk? - Where does your standard path force someone into exception handling? - What shared system has lost authority because stale material remains visible? - Which metric tells a clean story but cannot support a decision? - What would your team stop, retire, or refuse if strategy worked? ### Assessment options - **Concept map:** Trace purpose → operating choice → workflow → measure → review. - **Case memo:** Grade one book claim and propose a faithful repair. - **System audit:** Apply five elements and four lifecycle states to a live but low-risk service. - **Teach-back:** Explain why a repository is not a knowledge service. | Criterion | Strong work | |---|---| | Attribution | Separates author, third party, and analysis | | System reasoning | Names dependencies, owners, feedback, and exit | | Evidence | Matches claim strength to evidence strength | | Inclusion | Tests default paths, power, access, and labor | | Action | Uses a reversible experiment with measures and a stop rule | ### Facilitator cautions - City, transit, ecosystem, periodic-table, and machine metaphors can outrun the evidence. Return to actors, decisions, costs, harms, and exceptions. - Disclose that every source passage here is system-selected and no reader highlights were available. - Consent, workplace surveillance, layoffs, discrimination, and harmful research can be personally sensitive. Allow private reflection or opting out. - Do not present Towsey's models, this review's repairs, or current legal context as settled fact. - Never attribute the permanent-note candidates or posterior estimates to Towsey. ### Accessibility and multicultural adaptations - Provide the matrix as a table and plain-language checklist, not only a diagram. - Send terms and cases in advance; expand acronyms on first use. - Offer written, spoken, asynchronous, and small-group participation. - Do not require public disclosure of workplace harm or identity. - Replace large-technology cases with local public, nonprofit, open-source, or community examples. - Let learners revise categories and measures using their jurisdiction, language, infrastructure, and community governance. - Compensate community expertise when teaching moves into consultation or research. ### Further reading - [ResearchOps Community — About and eight pillars](https://researchops.community/about/) - [ICC/ESOMAR International Code, 2025](https://standards.esomar.org/assets/documents/icc-esomar-code-2025.pdf) - [W3C Web Content Accessibility Guidelines 2.2](https://www.w3.org/TR/WCAG22/) - Richard Rumelt, *Good Strategy/Bad Strategy* - Etienne Wenger, Richard McDermott, and William Snyder, *Cultivating Communities of Practice* - Steve Portigal, *Interviewing Users*, 2nd edition - `[[4. Archives/Book reviews/The Product is Docs]]` for a parallel operating-system treatment of documentation --- ## Prompt 12 — actions, PARA, experiments, and drift ### Action inventory #### Take - Define one research service's purpose, refusal rule, owner, users, quality threshold, and failure boundary. - Add maintenance, incident, correction, and retirement stages to any lifecycle plan. - Trace participant data and consent from collection through deletion and remedy. - Test one knowledge or support journey with a person who did not build it. - Tie every reported metric to a decision, denominator, quality guardrail, owner, and review date. - Make private escalation available for ethics, accessibility, privacy, identity, or power-sensitive support. #### Consider - Use the eight elements as a dependency prompt, then add local security, accessibility, records, labor, and decommissioning checks. - Create a light knowledge service before buying a research repository. - Co-design participant and knowledge taxonomies with affected users. - Publish collaboration and priority rights, including appeals and protected exploratory capacity. - Replace an individual “big picture” week with equitable team review time and coverage. #### Avoid - Buying a tool before defining the job, data path, support work, and exit. - Calling more studies, reports, or participants “scale” without quality and harm thresholds. - Treating synthetic participants as substitutes for people whose lives and contexts matter. - Using the book as legal, privacy, accessibility, or AI compliance guidance. - Turning study counts, repository views, or support tickets into worker-performance scores. - Implementing all eight elements at full enterprise depth for a small team. > [!danger] Most likely misuse > The most likely misuse is to turn the Planning Matrix into process theater: more forms, tools, dashboards, gates, and centralized control that make ResearchOps visible while research becomes slower, less accessible, less exploratory, or less accountable to participants. #### Questions needing more evidence - Which system changes improve participant outcomes and research quality in comparable settings? - Which operating measures predict better decisions rather than higher activity? - Which competency systems improve pay, promotion, retention, and belonging across groups? - Which AI tasks remain safe and valid with what review and consent controls? - Which parts of the model survive small-team and community-led implementation? ### PARA routing | Bucket | Material | Decision | |---|---|---| | Projects | None by default | No active implementation project was supplied; do not invent one | | Areas | Research practice, product discovery, DevRel, technical communication | Use only where ongoing responsibility already exists | | Resources | This review, vocabulary, claims, teaching pack, visual specifications | Durable reference material | | Archives | Source record and completed Work the Book run | Canonical location for the consolidated note | ### Reversible experiments All experiments involving other people should be disclosed as ordinary service-improvement conversations, never covert studies. | Experiment | Claim | Hypothesis and action | Measure | Duration | Stop condition | Review date | |---|---|---|---|---|---|---| | Research-request journey audit | C-003, C-016 | Map one low-risk request from need to decision; ask 3–5 recent users where it broke | Completion, handoffs, wait, missing decisions, accessibility barriers | 2 weeks | Stop if sensitive data or employment evaluation enters scope | 2026-09-17 | | Taxonomy exception log | C-009 | Record cases that do not fit current participant cohorts before changing categories | Exception count, affected groups, routing delay, user correction | 4 weeks | Stop if logging creates identity or privacy risk | 2026-10-01 | | Knowledge retrieval task test | C-010 | Ask 3–5 colleagues to find one current finding, its source, owner, and status | Success, time, wrong-version rate, confidence | 3 weeks | Stop if access controls could expose restricted data | 2026-09-24 | | Two-route support pilot | C-011 | Offer a shared routine channel plus private safety/access escalation | Resolution, wait, repeat questions, hidden labor, safe escalation use | 4 weeks | Stop if private cases appear in shared records | 2026-10-01 | | Metric-to-decision audit | C-015 | Review five recurring metrics and name the decision each changed | Metrics with owner, denominator, decision, guardrail, expiry | 2 weeks | Stop if the exercise turns into staff ranking | 2026-09-17 | | Protected review block | C-018 | Trial one shared, covered, two-hour system-review block rather than a leader-only week | Issues found, decisions made, roles able to attend, work displaced | 4 weeks | Stop if coverage failure harms service or shifts unpaid work | 2026-10-01 | ### Decision rules - **Adopt:** The experiment improves the chosen outcome without worsening quality, access, privacy, participant welfare, or hidden labor. - **Adapt:** Direction is positive but one group, step, or risk performs worse; change that part and rerun. - **Further test:** Signal is mixed, sample is thin, or an alternative explanation remains likely. - **Reject:** The practice adds process, surveillance, exclusion, or labor without outcome gain. Revise the related permanent-note candidate rather than preserving the rule. - **Defer:** Authority, safe data, staffing, or stakeholder consent is missing. Rejection mapping: taxonomy failure revises *A Participant Taxonomy Is a Governed Hypothesis*; tool-lifecycle failure revises *A Tool Purchase Is a Data, Labor, and Exit Policy*; dashboard failure revises *Operational Visibility Needs Quality and Harm Guardrails*; support failure revises *Support Gates Must Not Block Safety Access*; retrieval failure revises *A Knowledge Service Needs Lifecycle States, Not One Repository Status*. ### Drift monitoring | Topic | What could change | Monitor | Frequency | Reassess | Early trigger | |---|---|---|---|---|---| | AI research tools and synthetic users | Capability, errors, data use, pricing, disclosure | Primary vendor docs, task evaluations, regulator guidance | Quarterly | 2026-11-27 | Tool or policy change affecting participant data | | AI law | Scope, dates, duties, guidance | Relevant regulators; EU AI Act page where applicable | Before use and quarterly | 2026-11-27 | New jurisdiction, use case, or enforcement action | | Research ethics | Professional codes and emerging methods | ICC/ESOMAR, MRS, institutional ethics bodies | Semiannual | 2027-02-27 | Synthetic data, vulnerable population, or new collection method | | Privacy and participant rights | Local laws, retention, deletion, consent | Qualified local counsel and regulators | Before implementation | Per project | New country, data category, or vendor | | Accessibility | Standards, interpretations, tool conformance | W3C/WAI, user testing, procurement evidence | Annual and per tool | 2027-08-27 | Major tool or workflow change | | ResearchOps labor market | Roles, layoffs, team models, terminology | Transparent workforce reports | Annual | 2027-08-27 | Staffing or org-design decision | | Named vendors and surveys | Availability, methods, current data | Primary source and current methods | Before citation | On use | Link, product, or sample changes | **Low-drift claims:** Purpose should precede infrastructure; ongoing consent needs withdrawal and deletion routes; knowledge needs ownership and provenance; measures need decision links; priority rights should be explicit; maintenance and retirement need owners. Review these only when credible counterevidence or a changed application appears. --- ## Prompt 13 — visual thinking pack ### Visual 1 — book and operating-system map - **Purpose:** Show how the four conceptual chapters feed eight operational elements and renewal. - **Nodes:** Scale; strategy; system design; planning; recruitment; knowledge; support; tools; respect; money/metrics; priorities; talent; reflection. - **Connections:** Foundation flows into the operational ring; ethics and feedback constrain all elements; reflection returns to strategy. - **Grouping:** Foundation, infrastructure, governance, people/renewal. - **Uncertainty:** Dashed edge from operating system to “better outcomes” because comparative evidence is limited. - **Claims:** C-001–C-018. - **Source IDs:** EPUB-C01#sec3; EPUB-C02#sec6; EPUB-C03#sec4; EPUB-C04#sec8; EPUB-C05–C12; EPUB-CODA. - **Layout:** Left-to-right foundation; circular element cluster; feedback arrow beneath. ```mermaid flowchart LR SCALE[Scale] --> STRAT[Research + ResearchOps strategy] STRAT --> DESIGN[Operating-model design] DESIGN --> PLAN[Eight elements × lifecycle] PLAN --> REC[Recruitment] PLAN --> KM[Knowledge] PLAN --> SUP[Support] PLAN --> TOOL[Tools] PLAN --> ETH[Respect] PLAN --> MET[Money + metrics] PLAN --> PRI[Priorities] PLAN --> TAL[Talent] REC & KM & SUP & TOOL & ETH & MET & PRI & TAL --> SYS[Research operating system] SYS -. "outcome evidence incomplete" .-> OUT[Useful, ethical, sustainable research] OUT --> REF[Protected review] REF --> STRAT ``` ### Visual 2 — claim dependency and weakest link - **Purpose:** Show load-bearing claims and where evidence is weakest. - **Nodes and connections:** C-001/C-002 → C-003 → C-004 → C-006 → C-007/C-008 → practical elements → outcome. - **Grouping:** Definitions, strategy, design, implementation, outcome. - **Labels:** Dashed outlines mark claims supported mainly by cases and logic. - **Uncertainty:** C-002, C-003, C-006, and C-015 remain the main inferential gaps. - **Claims and sources:** C-001–C-018; E-001–E-017. - **Layout:** Top-down dependency tree. - **Implementation:** Use the Mermaid dependency graph in Prompt 4; preserve the `weak` dashed class. ### Visual 3 — evidence map - **Purpose:** Prevent practitioner cases, third-party sources, and current context from blending. - **Nodes:** Author experience; worked models; independent theory; selected company cases; vendor surveys; current standards; conclusions. - **Connections:** Each lane feeds claim grades; no lane alone reaches A-grade confidence. - **Grouping:** Source-origin columns. - **Uncertainty:** Vendor and company-case edges are dashed; current standards support context, not the book's effectiveness. - **Claims:** C-001–C-018. - **Source IDs:** E-001–E-017. - **Layout:** Three inputs converging on a judgment gate. ```mermaid flowchart LR A[Author experience and models] --> J{Claim judgment} T[Independent theory and codes] --> J C[Selected company cases] -.-> J V[Vendor surveys] -.-> J N[2025–26 current context] --> CTX[Current applicability] J --> B[8 B grades] J --> CC[9 C grades] J --> D[1 D grade] CTX --> LIMIT[Legal, AI, accessibility limits] ``` ### Visual 4 — scale feedback loop - **Purpose:** Show how demand can improve or degrade the system. - **Nodes:** Demand; operational load; standard paths; support; reuse; trust; research value; new demand; hidden labor; quality/access risk. - **Connections:** Reinforcing value loop and balancing risk loop. - **Grouping:** Desired loop in blue/green; failure loop in orange/red. - **Uncertainty:** The causal links are Towsey's mechanism plus analysis, not quantified effects. - **Claims:** C-001–C-003, C-009–C-016. - **Source IDs:** EPUB-C01#sec3–#sec11; EPUB-C05–C11. - **Layout:** Two interlocking loops. ```mermaid flowchart LR D[Research demand] --> L[Operational load] L --> S[Designed standard paths] S --> R[Reuse and support] R --> T[Trust and visible value] T --> D L --> H[Hidden labor and shortcuts] H --> Q[Quality, access, privacy risk] Q --> T REV[Review, repair, retirement] --> S Q --> REV ``` ### Visual 5 — stakeholder and power map - **Purpose:** Show who supplies, governs, funds, performs, and bears risk. - **Nodes:** Participants; researchers/PWDR; ResearchOps; leadership; finance/procurement; legal/privacy/security; HR; vendors; research consumers. - **Connections:** Data, money, support, decisions, employment power, and accountability. - **Grouping:** People researched; people doing research; internal governors; external providers. - **Uncertainty:** Participant appeal and co-governance links are often missing in the book; draw them dashed. - **Claims:** C-009–C-017. - **Source IDs:** EPUB-C05–C12. - **Layout:** Participants at center; actors around them; power arrows labeled. ```mermaid flowchart TD P((Participants)) -->|experience + data| R[Researchers / PWDR] OPS[ResearchOps] -->|routes + support| R LEAD[Leadership] -->|priorities + authority| OPS FIN[Finance / procurement] -->|budget + contracts| OPS GOV[Legal / privacy / security] -->|constraints + review| OPS HR[HR / people partners] -->|employment systems| R VEN[Vendors] -->|tools + services| OPS R -->|findings| CON[Research consumers] CON -->|decisions + demand| LEAD P -. "appeal / co-governance often unspecified" .-> OPS OPS -->|consent, care, remedy| P ``` ### Visual 6 — decision and action tree - **Purpose:** Decide whether to build, adapt, defer, or reject an operational change. - **Nodes:** Defined purpose; risk; existing system; small pilot; outcome/guardrail; adopt/adapt/reject/defer. - **Connections:** Yes/no decisions with a stop path for unsafe or unauthorized work. - **Grouping:** Before, pilot, review. - **Uncertainty:** Local authority and evidence decide the branch; the book cannot supply them. - **Claims:** C-004, C-008, C-014–C-016. - **Source IDs:** EPUB-C02#sec6; EPUB-C04#sec8; EPUB-C09–C11. - **Layout:** Top-down decision tree. ```mermaid flowchart TD A{Purpose and refusal rule defined?} -->|No| DEF[Define before building] A -->|Yes| B{Safety, privacy, access, authority clear?} B -->|No| DEFER[Defer and obtain qualified review] B -->|Yes| C{Can an existing path be repaired?} C -->|Yes| REPAIR[Run small repair pilot] C -->|No| PILOT[Run smallest new pilot] REPAIR --> D{Outcome improves without guardrail harm?} PILOT --> D D -->|Yes| ADOPT[Adopt with owner, review, and exit] D -->|Mixed| ADAPT[Adapt and retest] D -->|No| REJECT[Reject and record why] ``` ### Excalidraw-ready layout Create a landscape canvas with four horizontal bands. Band 1 contains four foundation cards: scale, strategy, operating model, planning matrix. Band 2 contains eight equal element cards in a ring. Band 3 places participants at center with researchers, ResearchOps, leaders, partners, and vendors around them. Band 4 contains measure, review, incident, repair, and retirement cards joined as a feedback rail. Use solid arrows for explicit author dependencies, dashed arrows for analysis extensions or weak evidence, and a red outline for participant-harm and surveillance risks. Put source IDs in each card's lower-right corner. ### Visuals omitted or represented elsewhere - **Standalone timeline:** Omitted because this is not primarily a historical argument; the context section supplies the few dates that matter. - **Standalone terminology map:** Omitted because the controlled-vocabulary table better preserves competing and operational definitions. - **Standalone comparison matrix:** Represented by the operating-model terms, multicultural-transfer table, and action decision rules. - **Standalone concept map:** The book map already performs this role. - **Standalone action map:** The decision tree and experiments table perform this role. --- ## Final synthesis ### The book in one sentence Research scales responsibly when an organization designs and maintains the surrounding system—purpose, people, services, data, tools, controls, money, decisions, and learning—without trying to mass-produce human judgment. ### What changed in the priors - Confidence increased that lifecycle planning, stewardship, and explicit priority contracts are useful operating disciplines. - Confidence remained moderate that the eight-element taxonomy is complete, that tool changes cause culture change, and that operational metrics improve outcomes. - The book strengthened the link between recruitment logistics and participant ethics. - The RKM chapter reinforced the need to separate social knowledge creation from governed durable knowledge while adding lifecycle states. - The AI chapter aged quickly; its durable contribution is tool-governance discipline, not a 2023 capability verdict. - Protected reflection remains plausible but under-evidenced and unequally available. ### Best next move Do not build a comprehensive ResearchOps program from the book. Pick one live, low-risk research journey and run the two-week journey audit. That will reveal whether the first need is purpose, ownership, access, support, knowledge, privacy, tooling, or retirement—and it creates local evidence before process expands. --- ## Session completion check ### Prompt status | Prompt | Status | Usable output | |---:|---|---| | 0 | Executed with Layer 1 substitution | Source record, Readwise check, coverage, anchors, index, clusters, defects | | 1 | Executed with 🔶 inferred inputs | YAML, contract, guardrails, 27 questions, 12 priors, drift | | 2 | Executed | Full book map, central project, summary, hotspots, questions, strategy | | 3 | Executed | Classification, 21-term vocabulary, contested language, category errors | | 4 | Executed | 18 claims, source labels, evidence expectations, falsifiers, graph, question map | | 5 | Executed | Dependence-aware policy, 12 updates, warnings, fallback | | 6 | Executed | 12 chapter slots plus coda, spine, 17-item ledger, reading plan, criteria | | 7 | Executed | Reconstruction, 18 evaluations, grade distribution, bias scan, counterevidence, repairs | | 8 | Executed | Historical, institutional, economic, technological, and current context; lineage map | | 9 | Executed | Inclusion, omission, transfer, EPUB accessibility, reading demand, supplements | | 10 | Executed under one-file constraint | Fleeting/literature notes, six permanent candidates, structure decisions, attribution audit | | 11 | Executed | Audience, objectives, sequence, questions, cases, assessment, adaptations, cautions | | 12 | Executed | Actions, PARA, six experiments, decision rules, drift monitor | | 13 | Executed | Six visual specifications, Mermaid, Excalidraw layout, omission reasons | ### Partial failures and adaptations 1. **Prompt 0:** No exact Readwise Reader document or highlight record existed. The complete local EPUB became Layer 1; reader emphasis remains unavailable. 2. **Prompt 0:** Pandoc did not preserve chapter-local footnote links reliably in merged output. Original XHTML controlled evidence and passed validation. 3. **Prompt 10:** The canonical prompt normally splits permanent notes and adds reciprocal MOC links. The explicit one-file contract takes priority, so candidates and intended destinations remain embedded. No adjacent file was created or edited. 4. **Prompt 13:** A timeline and duplicate terminology, concept, comparison, and action diagrams were intentionally omitted because existing tables and visuals communicate them better. ### Missing material ranked by consequence 1. **High relevance consequence:** Reader purpose, beliefs, constraints, notes, and highlights. Their absence limits personalization but not source coverage. 2. **High evidential consequence:** Comparative outcomes, failed implementations, denominators, participant outcomes, and raw organizational data. 3. **High implementation consequence:** A named organization, live research journey, jurisdiction, authority structure, and risk classification. 4. **Moderate current-context consequence:** Current tool-specific tests and qualified legal/privacy review. 5. **Moderate accessibility consequence:** Screen-reader, keyboard, contrast, zoom, and cognitive-access testing of the EPUB. 6. **Low bibliographic consequence:** Independent verification of every one of the EPUB's 146 notes; only selected current context was checked. ### Assumptions made - 🔶 The purpose is practical evaluation for research-heavy product, DevRel, technical communication, and knowledge work. - 🔶 Analytical depth and Archives routing fit the request and existing review collection. - 🔶 Participant dignity, disability access, multicultural transfer, power, and labor sustainability are the DEIB lens. - Existing vault notes establish adjacency, not Courtney's beliefs about this book. - A rating of 4/5 reflects handbook usefulness minus evidence and transfer limits. ### Contradictions resolved - The one-file request overrides Prompt 0 and Prompt 10's default separate-note outputs. - “All scalable systems are efficient/sustainable” is narrowed to declared thresholds under tested demand. - The eight elements are treated as a dependency lens, not a complete ontology. - “Human in the loop” is treated as necessary but insufficient for AI governance. - The EPUB's arithmetic, title, compounding, and gravitational-constant defects are recorded rather than silently corrected. ### Final state - Executed prompts: `[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13]` - Skipped prompts: `[]` - Remaining prompts: `[]` - Next prompt: `null` - Session status: `complete_with_documented_source_and_reader_context_limits` - Attribution audit: passed after source/analysis separation - Separate Reader publication: not requested and not performed - Separate MOC, literature-note, or permanent-note edits: not requested and not performed ### Session record ```yaml title: "Work the Book — Prompts Used This Session" purpose: "Record the exact prompts executed during this Work the Book session so they can be reviewed, revised, or rerun." book_title: "Research That Scales: The Research Operations Handbook" book_author: "Kate Towsey" book_year: 2024 source_of_truth: "/Users/courtneyrobertson/Documents/2nd Brain/6. Templates/Books/Work the Book.md" working_note: "/Users/courtneyrobertson/Documents/2nd Brain/4. Archives/Book reviews/Research That Scales.md" executed_prompts: [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13] skipped_prompts: [] remaining_prompts: [] session_date: "2026-08-27" session_status: "complete_with_documented_source_and_reader_context_limits" ``` ## Sources ### Primary source - Kate Towsey, *Research That Scales: The Research Operations Handbook*. Rosenfeld Media, 2024. Supplied local EPUB; SHA-256 recorded in front matter. ### Current publication and context checks - [Rosenfeld Media — Research That Scales](https://rosenfeldmedia.com/books/research-that-scales/) - [ResearchOps Community — About ResearchOps and the eight pillars](https://researchops.community/about/) - [ICC/ESOMAR International Code on Market, Opinion and Social Research and Data Analytics, 2025](https://standards.esomar.org/assets/documents/icc-esomar-code-2025.pdf) - [European Commission — AI Act implementation and timeline](https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai) - [W3C — Web Content Accessibility Guidelines 2.2](https://www.w3.org/TR/WCAG22/) ### Vault context consulted - `[[3. Resources/MOCs/Book Review Synthesis MOC]]` - `[[3. Resources/MOCs/Product & Startup Instincts MOC]]` - `[[3. Resources/MOCs/Personal Knowledge Management MOC]]` - Existing permanent notes linked in “Why this book belongs in this vault” and Prompt 10.