Knowledge, expertise, and professional autonomy

Residents knew the official account did not fit their water

In April 2014, while Flint was under a state-appointed emergency manager, the city began drawing drinking water from the Flint River. The treatment plant did not maintain the corrosion control required after the source change, and the result prolonged residents' exposure to lead. State and federal reviews also recorded repeated water-quality complaints, bacterial violations, disinfection byproduct violations, high household lead readings, and official delays in recognizing the combined risk.¹

The U.S. Environmental Protection Agency's Region 5 received 87 citizen complaints between May 2014 and its January 2016 emergency order. Complaints described color and odor before increasingly specific concerns about microbes, disinfection byproducts, Legionella, and lead. The inspector general found that Region 5 had no system for cataloguing the complaints or using their volume as an incident indicator; responses sometimes referred residents back to the state or city whose account they were contesting.²

Residents therefore had repeated, situated observations but did not control the sampling protocol, regulatory interpretation, laboratory, or authority to order protective action. Their reports were neither a substitute for chemical and clinical testing nor mere noise. They identified a system whose official representation no longer fit lived conditions.

The Flint Water Advisory Task Force called the crisis a series of government failures marked by intransigence, delay, inaction, and environmental injustice. It also credited engaged residents and people inside and outside government who challenged official assurances. The task force was commissioned by the governor and worked from interviews and the public record rather than subpoena power, so its findings are an official investigative judgment with stated access and time-boundary limits—not a resident-controlled account or a judicial finding.¹

A resident-led coalition made another account inspectable

LeeAnne Walters collaborated with EPA scientist Miguel Del Toral and later with Marc Edwards's Virginia Tech team after high lead readings appeared at her home. Del Toral's June 2015 memorandum joined household results to the absence of corrosion control and weaknesses in the official sampling approach. In August, the Virginia Tech–resident collaboration supplied kits and laboratory support for a geographically distributed sampling effort involving 269 homes.³

Different participants contributed different forms of warrant. Walters and other residents supplied local access, repeated observation, organizing, and labor. Del Toral connected those observations to the Lead and Copper Rule and agency evidence. Engineers connected water chemistry, pipe corrosion, sampling practice, and laboratory results. The coalition did not make every participant equally powerful, but it created a chain in which the observation, method, sample, analysis, and institutional response could be inspected.³

That sequence resists an expert-rescue story. Residents were investigators and organizers before outside credentials increased the evidence's mobility. Professional autonomy served a public purpose when it joined resident knowledge to testable methods; the same sequence reveals how much unpaid work and exposure were required before the warning acquired standing.

Clinical evidence arrived after children had already borne exposure

Mona Hanna-Attisha and colleagues compared blood-lead records for children younger than five before and after the source change. In their sample, the percentage of Flint children with elevated blood-lead levels increased from 2.4% to 4.9%; high-water-lead areas increased from 4.0% to 10.6%, while the change outside Flint was not statistically significant. Their spatial analysis also located the greatest increases in socioeconomically disadvantaged neighborhoods.⁴

The study made population-level change visible in a form public-health agencies could evaluate. It did not identify the exposure or future outcome of every child, and its retrospective before-and-after design was constrained by who was tested, available addresses, residential movement, and area-level water data. Those limits narrow the inference; they do not turn the observed increase into evidence that no population harm occurred.⁴

The evidentiary timing is ethically central. Blood records gained authority only after children had become detectors for an infrastructure failure. A system that waits for clinical injury before accepting resident and technical warnings may produce strong retrospective evidence while failing at prevention.

Oversight and civil-rights inquiries answer different questions

The EPA inspector general reconstructed regulatory implementation, federal oversight, complaint handling, and emergency response. It found that Michigan did not maintain two Lead and Copper Rule requirements and that unclear roles, weak risk assessment, ineffective communication, and unused oversight tools delayed the federal response.⁵

The Michigan Civil Rights Commission held three hearings, took testimony from residents and experts, conducted field visits, and concluded that race and systemic racism played a significant role in producing the crisis. Its mandate was civil rights and structural discrimination rather than water chemistry or individual liability. The commission also acknowledged that it had not been proactive enough when Flint residents' voices were being ignored.⁶

These records overlap without becoming interchangeable. The task force assigns administrative responsibility; Del Toral documents a contemporaneous technical warning; the clinical study estimates a sampled population change; the EPA audit evaluates oversight; the civil-rights commission examines structural power; and the participant-authored citizen-science account reconstructs a coalition from inside it. Agreement across them strengthens some conclusions, but each source retains its mandate, standpoint, and limitations.

I10-N01: specialized knowledge is also a claim to jurisdiction

I10-N01 groups priesthoods, guilds, professions, and universities as different ways specialized knowledge has been formed, preserved, authorized, and contested. It is an analytical grouping, not one continuous European lineage or a claim that the institutions share the same authority, membership, or public obligations.

Andrew Abbott's comparative sociology treats modern professions as a system in which occupational groups compete for jurisdiction over expert work. His account links professional work to diagnosis, treatment, inference, academic knowledge, audiences, and settlements, while also examining exclusion, internal stratification, clients, and power.⁷ It supports asking who may define and perform expert work; it does not make credential closure inherently legitimate or describe every priesthood, guild, or university.

Flint makes the jurisdictional question concrete. Residents could observe and sample their water, but agencies and laboratories held different powers to classify the observation, validate the method, and authorize a response. The problem was not a choice between lived knowledge and technical method. It was whether a warning could cross jurisdictions without losing its origin or waiting for injury.

I10-N02: Weber joins technical qualification to office

Max Weber's bureaucracy analysis in Economy and Society describes fixed jurisdiction, hierarchy, written files, trained officials, career office, and technical qualification. He also connects bureaucratic power to specialized knowledge, information acquired in office, and official secrecy.⁸

Weber offers an ideal-type and historical analysis, not an evaluation of Flint. Applied as a later comparison, it shows how expertise attached to office can make administration continuous and reviewable while giving an office power over what becomes administratively real. A rule-bound sample can be durable evidence even when the rule or sample selection excludes the household condition that prompted the inquiry.

I10-N02 therefore does not imply that impersonality or expertise caused the crisis. It identifies a governance problem: technical independence from political whim needs routes by which affected people and independent specialists can challenge the official record.

I10-N03: Follett separates relevant authority from personal rank

Mary Parker Follett's “The Giving of Orders,” preserved in Dynamic Administration, argues for jointly studying a changing situation and deriving action from its requirements rather than personal command. Her “law of the situation” is a normative and practical proposal for depersonalizing orders, not evidence that joint inquiry removes hierarchy or conflict.⁹

No cited Flint record identifies Follett as a source for residents, engineers, clinicians, or officials. I10-N03 is a conceptual lens: water chemistry, household observation, pediatrics, regulation, and emergency authority could have been coordinated around the water residents actually received rather than around rank alone.

The situation does not define itself. People still select evidence, decide who participates, settle values that no specialist method can answer, and remain responsible for action. Follett makes functional authority visible without turning “the facts” into an escape from power or accountability.

I10-N04: Drucker makes contribution the test of knowledge work

Peter Drucker defines an executive broadly enough to include a knowledge worker whose position or knowledge requires decisions that affect organizational performance. In The Effective Executive, the relevant practices include recording time, asking what contribution is needed, making strengths productive, choosing priorities, testing disagreement, and turning a decision into action and feedback.¹⁰

Drucker's source is prescriptive and case-based; it does not measure how common professional autonomy is or whether his practices improve performance, equity, or well-being. I10-N04 consequently treats the knowledge worker as a responsible contributor, not as an empirically validated worker category whose expertise should outrank every other claim.

Flint sharpens the question “contribution to whom?” An expert can satisfy an office's procedure while affected families continue to bear the result. Del Toral, the Virginia Tech team, and Hanna-Attisha's group carried evidence across organizational boundaries. Their actions do not make credentialed dissent automatically correct; they show why autonomy must remain answerable to method, peer challenge, affected people, and consequences.³⁴

I10-N05: Simon makes expertise selective rather than omniscient

Herbert Simon described recognition of familiar patterns and highly selective search as central processes in problem solving. He tied both to domain knowledge and emphasized that a problem's representation can substantially change how easy it is to solve. His account also treats attention to one item as an opportunity cost that leaves others unattended.¹¹

This cognitive program does not reduce every expert judgment to an explicit rule or prove that Simon's models exhaust tacit, social, emotional, and embodied knowing. It does explain why a checklist can support expertise without reproducing it: a representation selects cues and paths, while learned knowledge helps distinguish promising searches from fruitless ones.

Simon did not study Flint in the cited work. I10-N05 applies his account as a later comparison: an official compliance representation narrowed search, while household samples, corrosion analysis, and clinical records changed which cues were salient and when inquiry could stop. The comparison is analytical, not a claim that Simon influenced any participant.

I10-N06: Nonaka models movement between tacit and explicit knowledge

Ikujiro Nonaka and Hirotaka Takeuchi describe socialization, externalization, combination, and internalization as interactions between tacit and explicit knowledge in The Knowledge-Creating Company. The theory joins those interactions to an organizational knowledge spiral.¹²

The model is a primary authorial proposition rather than proof that every innovation follows four modes. Stephen Gourlay's independent critique argues that the cases permit simpler explanations, that inherently tacit knowledge can drop out of the framework, and that managerial validation helps determine what counts as knowledge. The critique does not prove that no knowledge conversion occurs; it establishes a serious conceptual and evidentiary boundary.¹²

I10-N06 uses movement as a question, not a literal substance transfer. In Flint, place-specific observation became samples, maps, memoranda, and clinical analysis, then returned to organizing and public action. That movement depended on unequal access to laboratories, credentials, media, and administrative authority; codification did not erase the people or conditions that made the evidence possible.³

I10-N07: communities of practice make learning social and bounded

Etienne Wenger's theory of communities of practice treats engagement in social practice as central to learning and identity. The community, rather than only the individual or formal institution, becomes a unit for studying shared enterprise over time.¹³

The framework does not establish that every informal group learns well or serves people outside it. Later scholarly work collected under Beyond Communities of Practice specifically foregrounds language, power, conflict, and wider social context that can be overlooked in celebratory applications.¹³

I10-N07 therefore covers maintainers, clinicians, researchers, software practitioners, and other groups only as a comparative lens. Shared problems, artifacts, stories, and review can develop capability across formal boundaries; the same boundary can exclude newcomers, hide conflicts, or discount affected people who are not recognized as practitioners.

I10-N08: generative AI raises a provenance and authority problem

The National Institute of Standards and Technology identifies confabulation as confidently presented false or erroneous generated content. Its generative-AI risk profile recommends documenting knowledge limits and upstream data sources, fact-checking generated information, reviewing citations, recording human oversight roles, and tracking content provenance and its limitations.¹⁴

NIST's profile is a voluntary risk-management framework, not a measured claim that every system fails in the same way or that provenance techniques make an output true. I10-N08 is correspondingly provisional. Generated synthesis may help people explore a corpus, but fluent output does not inherit the situated experience, professional duty, consent, or accountability of the people and institutions whose language it recombines.

A Flint-like dispute makes the risk legible. A corpus dominated by official records could reproduce the agency account more fluently than resident knowledge that was never digitized or categorized. The safeguard is not to treat testimony as an untested technical claim; it is to preserve who observed what, how it was tested, what remains disputed, which institution acted, and who is responsible if the output informs a harmful decision.¹⁴

The seven edges are provisional comparisons, not a genealogy

The node sequence organizes questions about authority and knowledge. Every edge is explicitly comparative; none establishes direct historical transmission or a universal causal mechanism.

• I10-E01 is a theoretical-codification comparison, Grade C: it places the editorial grouping in I10-N01 beside Weber's account of technically qualified office in I10-N02.⁷⁸
• I10-E02 is a reaction-or-critique comparison, Grade B: Follett's situation-derived authority in I10-N03 is read as a relational alternative to rank-centered bureaucracy, not as a documented response to Weber.⁸⁹
• I10-E03 is a managerial-generalization comparison, Grade B: Drucker's responsible knowledge worker in I10-N04 broadens the question of functional contribution; no cited record establishes a Follett-to-Drucker transmission for this claim.⁹¹⁰
• I10-E04 is a theoretical-codification comparison, Grade C: Simon's recognition, representation, and selective search in I10-N05 provide a cognitive account alongside Drucker's prescriptive knowledge-worker category, not a measured derivation from it.¹⁰¹¹
• I10-E05 is a theoretical-codification comparison, Grade C: Nonaka's tacit–explicit model in I10-N06 shifts the unit from individual problem solving to organizational knowledge creation; the sources do not establish a Simon-to-Nonaka lineage.¹¹¹²
• I10-E06 is a managerial-generalization comparison, Grade B: communities of practice in I10-N07 extend inquiry from a company knowledge model to learning and identity across formal boundaries, while retaining power and exclusion as open questions.¹²¹³
• I10-E07 is an enabling-technology comparison, Grade C: I10-N08 asks how generated synthesis changes provenance, verification, and authority around codified knowledge; it is a current research question, not an estimated productivity or learning effect.¹³¹⁴

All seven relation records retain provisional status and empty internal source arrays. The adjacent external notes support the node descriptions and the stated comparison boundaries; they do not promote the edges into demonstrated influence. Empty counterevidence and reviewer arrays mean that neither search nor peer review is complete.

Related paths are typed as lenses, not evidence shortcuts

The six related paths carry distinct navigation roles.

• Dynamic Administration, Administrative Behavior, and The Sciences of the Artificial are primary-work lenses for Follett's and Simon's arguments. They do not document influence on the Flint actors.
• Organizational ignorance and culture, informal organization, trust, and voice are adjacent-idea comparisons about whether a signal can be noticed, spoken, and authorized. The links assert no shared empirical mechanism.
• Benefit for all life is an ethical lens on who bears consequences and whose knowledge enters a decision. It is not a historical source or a measured impact claim.

Sixty-nine institution paths are comparative prompts

Every institution path below is typed analytical comparison. The groupings do not assert equivalence, lineage, common mechanism, relative success, transferability, or endorsement. Evidence for each case remains with its own institution record.

• Pre-Constantinian church, British Royal Navy, Prussian–German General Staff, and DuPont prompt questions about formation, office, craft, command, recordkeeping, and specialized jurisdiction across unlike periods.
• IBM under Thomas J. Watson Sr. and Jr., U.S. Navy nuclear propulsion and submarine command, 3M, Boeing, Microsoft, Netscape, Novo Nordisk, NVIDIA, and Zeiss prompt comparisons among technical specialization, operating knowledge, hierarchy, finance, review, and product authority.
• ACOFOP community forest concessions, Aravind Eye Care System, Aswan High Dam administration, Botswana's diamond developmental state, Brazil's Family Health Strategy in SUS, Cairo's Zabaleen recycling system, Cameroon's community-forestry system, Canadian potlatch ban, Colombian Coffee Growers Federation and Cenicafé, Columbia River Inter-Tribal Fish Commission, Eskom under state capture, and Ethiopian Airlines prompt questions about local knowledge, professional systems, development authority, extraction, and accountability to communities.
• Popular committees of the First Intifada, Fujimori's reproductive-health program and forced sterilization, Ghana's Community-based Health Planning and Services, Guatemala's termination of CICIG, Haiti's 2010 humanitarian cluster response, Imperial Chinese examination system, Iran's Behvarz primary health network, ITRI–TSMC semiconductor ecosystem, Japanese quality-control circles, Karuk cultural fire institutions, and Te Kōhanga Reo prompt comparisons among credential, field knowledge, cultural authority, participation, and the power to reject an imposed expert account.
• Madagascar's GELOSE resource-management transfers, Mali Office du Niger, Menominee Tribal Enterprises, MINUSTAH and Haiti's cholera response, Morocco's women's argan cooperatives, Namibia's communal conservancies, Navajo livestock reduction, Navajo Nation Peacemaking, Niger's farmer-managed natural regeneration, Northern Nigeria polio vaccine boycott and legitimacy repair, Northern Rangelands Trust and Kenya's community conservancies, and Oman's aflaj irrigation institutions prompt questions about Indigenous, local, gendered, clinical, ecological, and administrative knowledge without collapsing those forms into one category.
• Operation Condor, PDVSA's institutional hollowing, South Africa's Treatment Action Campaign, South Africa's Truth and Reconciliation Commission, Sudan Gezira Scheme, Ushahidi, UZACHI community forestry, and Vale and the Brumadinho dam disaster prompt comparisons among secrecy, repression, institutional loss, public mobilization, testimony, distributed reporting, professional judgment, and post-harm accountability.

Four Workloop propositions remain research hypotheses

I10-P01 proposes attaching provenance, confidence, review status, and domain authority to durable knowledge artifacts. Promotion requires evidence that people can retrieve, assess, correct, and reuse artifacts more reliably without hardening credentialism, exposing confidential sources, or making obsolete authority harder to challenge.

I10-P02 proposes using AI as a research and synthesis participant whose claims remain inspectable and attributable. Blinded evaluation must compare claim accuracy, source fidelity, omission, correction cost, and inspectability. Fluent prose that launders weak evidence, displaces responsibility, or hides labor, data, and resource costs fails the proposition.¹⁴

I10-P03 proposes expert escalation and peer-review routes that do not require every concern to travel through management hierarchy. Evaluation must compare response time, error interception, access to relevant expertise, conflicts of interest, exclusion, and retaliation—not merely count escalations.

I10-P04 proposes treating mentoring and capability development as work products rather than invisible overhead. Evidence must track learner independence, capability transfer, resilience, and mentor workload while testing whether formalization coerces care, quantifies relationship poorly, or penalizes mentors for outcomes beyond their control.

The four propositions are tests, not findings attributed to Flint participants, Weber, Follett, Drucker, Simon, Nonaka, Wenger, NIST, or the institution paths. No organizational profile or structured impact record is encoded because the cited evidence does not support one comparable estimate of benefit or harm across the conceptual sequence and sixty-nine cases. That absence is an evidence boundary, not an implication that unmeasured effects are zero.

Expertise is a route by which reality can contest office

Flint does not settle the conflict between professional discretion and public control. Managers cannot know every specialty, experts can be mistaken or captured, and lived observation does not answer every causal question. A more defensible route preserves observation, method, provenance, disagreement, independent review, authority to act, and appeal.

“The experts said” is inadequate because it hides which expert, working for whom, within which jurisdiction, using which evidence, and answerable to whom. Professional autonomy becomes legitimate when it helps reality contest an official account before the people who supplied the warning must also supply the injury.

Additional reciprocal institution comparisons

Newly developed institutional records add these reciprocal comparison paths:

• Apple — institution-comparison
• Google / Alphabet — institution-comparison
• Linux ecosystem — institution-comparison
• Meta / Facebook — institution-comparison
• Purdue Pharma — institution-comparison
• Renaissance Italy — institution-comparison
• Steward Health Care — institution-comparison
• Vanguard — institution-comparison
• Wikipedia — institution-comparison

Each path identifies a sourced case where this idea is a defining emphasis. The relation is editorial comparison, not evidence of direct influence, shared terminology, or equivalent outcomes.