CounterpediaKnowledge you can trace back to the source.

/research/demo-corpus-twelve/flint-water-crisis-demo-corpus-dossier-v0-1-linked

Research artifactnot admittedlinked prototypedossier

FLINT WATER CRISIS DEMO CORPUS DOSSIER

Status: RESEARCH / PRE-CAPTURE ONLY

source: COUNTERPEDIA_DEMO_CORPUS_TWELVE_v0_1/FLINT_WATER_CRISIS_DEMO_CORPUS_DOSSIER_v0_1.md
sha256: fbf7f04ae79512e3513dc5efd43aada116b85409c891bccec8075ab5a6fcc845

Status: RESEARCH / PRE-CAPTURE ONLY
Subject: Flint water crisis — source-water switch, corrosion control, lead sampling, blood-lead evidence, emergency response, remediation, and current system status
Scope: source inventory, authority/posture, record-contradiction map, provisional five-claim matrix, capture hazards, reverse-source candidates
Explicit non-claims: This dossier creates no CaptureReceipt, SourceEdition, SRS receipt, custody, admission, final standing, article prose, or canonical Counterpedia identity.


A. SUBJECT / DISAMBIGUATION

Common name

Flint water crisis

Exact event scope

The City of Flint, Michigan's April 2014 switch from Detroit-supplied Lake Huron water to the Flint River; treatment and corrosion-control decisions; lead release from service lines/plumbing; official and independent sampling; elevated blood lead evidence among children; state/federal emergency response; service-line replacement and treatment changes; and the later distinction between system-level regulatory compliance and household-level lead risk.

This dossier is not a general history of Flint, all contaminants associated with the crisis, every criminal/civil proceeding, or every present-day public-health question.

Important identifiers / terms

Naming / semantic traps

  1. "Flint's water tested at X ppb" is usually incomplete. X may be an individual sample, a household first-draw value, a sequential sample, a 90th percentile statistic, a system compliance value, or a research-study summary.

  2. The LCR action level is not a health-based "safe tap" threshold. EPA describes it as a treatment-technique trigger/measure of corrosion-control effectiveness. EPA's MCLG for lead is zero.

  3. System compliance ≠ every home below the action level. A 90th percentile of 6 ppb can coexist with individual samples above 12 or 15 ppb.

  4. An independent 90th percentile and an official 90th percentile can differ without one being mathematically fraudulent. Sampling frame, site selection, stagnation, flushing, bottle geometry, timing, and handling matter.

  5. 2015 "below 15 ppb" official 90th-percentile results did not mean no lead problem existed. EPA OIG later found the sampling program/site inventory and corrosion-control response were defective.

  6. "Flint water is safe now" is temporally and denominator ambiguous. It may mean the distribution system meets current LCR compliance, not that every premise, fixture, service line, or exposure risk is lead-free.

  7. "Flint water is still unsafe" can be equally overbroad if meant as a current system-level regulatory claim. Current data show continuing LCR compliance.

  8. Blood lead prevalence values are study-specific. Hanna-Attisha and CDC analyses use different age bands, time windows, geographic methods, and statistical models.

  9. EPA OIG findings are Inspector General findings, not automatically the final institutional position of every EPA office.

  10. A settlement requiring service-line replacement does not itself equal an admission of liability. The 2017 agreement expressly states it was without admission of any issue of fact or law.

  11. A current webpage label is not automatically trustworthy document metadata. EPA's current Flint documents page appears to label the termination of its emergency order as "September 1, 2021," while the linked PDF and a May 2025 EPA announcement indicate a materially later termination action. Preserve the metadata conflict rather than silently correcting it.

Temporal bounds

Core crisis: April 2014–2017.
Regulatory/remediation arc: 2016–2025.
Current measurement/status: through August 9, 2026, with current data treated as time-sensitive and subject to recapture before publication.

Why fertile for Counterpedia

Flint is the strongest Demo Corpus subject so far for temporal truth + measurement discipline.

The page can truthfully contain all of these:

  • official 2015 system 90th-percentile values below the then-federal 15 ppb action level;

  • independent 2015 sampling with a substantially higher 90th percentile;

  • later official findings that the sampling/site-selection and corrosion-control program was defective;

  • elevated blood-lead evidence;

  • later system-level compliance;

  • current individual samples that still exceed Michigan's 12 ppb action level;

  • an EPA emergency order that was later terminated.

The value is not deciding which one number is "the truth."

It is preserving: sample population

  • sampling protocol

  • statistic

  • date

  • regulatory meaning

  • household/system scope

  • source posture.


B. SOURCE INVENTORY

FL-S01 — EPA OIG 2018 Management Weaknesses report

Exact title: Management Weaknesses Delayed Response to Flint Water Crisis
Issuer: U.S. Environmental Protection Agency, Office of Inspector General
Report: 18-P-0221
Date: July 19, 2018
Landing: https://www.epa.gov/office-inspector-general/report-management-weaknesses-delayed-response-flint-water-crisis
Exact PDF: https://www.epa.gov/sites/default/files/2018-07/documents/_epaoig_20180719-18-p-0221.pdf

Format: PDF
Authority / posture: EPA OIG audit/evaluation finding. OIG explicitly notes its report represents OIG opinion and does not necessarily represent final EPA position.

May support

  • MDEQ misinterpretation/implementation of LCR requirements;

  • Flint's failure to maintain a required service-line inventory;

  • absence of continuous corrosion control after the April 2014 source switch;

  • delayed EPA response and Region 5 oversight/communication weaknesses;

  • official 2015 LCR sampling results of 6 ppb and 11 ppb at the 90th percentile;

  • OIG conclusion that damaged infrastructure and lead release persisted until later corrective treatment.

Cannot establish

  • every individual exposure;

  • criminal liability;

  • final agency-wide EPA legal position.

Captureability: HIGH
Priority: P0


FL-S02 — EPA OIG 2022 follow-up

Exact title: EPA Needs to Complete Implementation of Remaining Recommendations to Improve Its Response to Drinking Water Contamination
Issuer: EPA OIG
Report: 22-P-0046
Date: May 17, 2022
Landing: https://www.epa.gov/office-inspector-general/report-epa-needs-complete-implementation-remaining-recommendations-improve
Exact PDF: available from the landing page

Authority / posture: Inspector General follow-up audit.

May support

  • EPA had certified all nine corrective actions complete;

  • OIG later found three corrective actions did not fully address the original deficiencies;

  • distinction between management closure and later independent audit sufficiency.

Counterpedia value A control can be marked "complete" in one administrative system yet later be judged insufficient by an oversight body.

Captureability: HIGH
Priority: P1


FL-S03 — EPA January 21, 2016 SDWA §1431 Emergency Administrative Order

Exact title: Emergency Administrative Order under Safe Drinking Water Act §1431 concerning Flint
Issuer: U.S. EPA
Date: January 21, 2016
Exact PDF: https://www.epa.gov/sites/default/files/2016-01/documents/1_21_sdwa_1431_emergency_admin_order_012116.pdf

Format: PDF, 18 pages
Authority / posture: federal emergency administrative order with mandatory/protective requirements.

May support

  • EPA's formal emergency determination;

  • finding that state/city actions were inadequate to protect public health at that time;

  • ordered treatment, monitoring, sampling, technical and public-information actions.

Cannot establish

  • all crisis causation;

  • later current system status.

Captureability: HIGH
Priority: P0


FL-S04 — EPA termination of Flint emergency order

Current EPA documents page:
https://www.epa.gov/flint/flint-drinking-water-documents

Linked PDF current path:
https://www.epa.gov/system/files/documents/2025-05/flint-termination-letter-1431-er-order.pdf

Contemporaneous EPA announcement: May 19, 2025 EPA announcement lifting the emergency order after requirements were completed.

Authority / posture: EPA termination action / current agency status.

Critical metadata anomaly The current Flint documents page appears to label the linked termination item as September 1, 2021, while:

  • the linked object lives under a 2025 path;

  • the PDF refers to a 2024 administrative compliance order;

  • EPA publicly announced termination on May 19, 2025.

Counterpedia rule Do not silently "correct" the webpage label. Capture:

  • page observation;

  • linked PDF bytes;

  • source-owned dates inside the PDF;

  • May 19, 2025 release;

then adjudicate the metadata discrepancy.

Captureability: HIGH
Priority: P0


FL-S05 — EPA February 29, 2016 LCR sampling clarification memorandum

Exact title: EPA Lead and Copper Rule sampling memorandum dated February 29, 2016
Issuer: U.S. EPA
Exact PDF: https://www.epa.gov/sites/default/files/2016-02/documents/epa_lcr_sampling_memorandum_dated_february_29_2016_508.pdf

Format: PDF
Authority / posture: official EPA technical/regulatory guidance/clarification.

May support

  • removing/cleaning faucet aerators before sampling can mask particulate lead;

  • pre-stagnation flushing can lower sampled lead and bias results;

  • bottle configuration/sampling protocol can affect lead measurements;

  • need for protocol consistency and risk-sensitive sampling.

Counterpedia value This is the key source proving that sampling method is part of the fact.

Captureability: HIGH
Priority: P0


FL-S06 — EPA 2016 review of MDEQ drinking-water program

Exact title: Review of the Michigan Department of Environmental Quality Drinking Water Program 2016
Issuer: U.S. EPA
Exact official surface: available through EPA Flint/drinking-water documents and EPA archived program-review material.

Authority / posture: federal program review of state primacy/implementation.

May support

  • LCR site-selection and tier-classification problems;

  • contradictory site status;

  • excluded/invalidated samples;

  • program oversight concerns.

Cannot establish

  • every excluded sample was intentionally manipulated;

  • criminal intent.

Captureability: HIGH
Priority: P1


FL-S07 — Flint Water Advisory Task Force Final Report

Exact title: Flint Water Advisory Task Force Final Report
Issuer/body: independent task force appointed by Michigan Governor Rick Snyder
Date: March 21, 2016
Exact PDF:
https://www.michigan.gov/flintwater/-/media/Project/Websites/formergovernors/Folder6/FWATF_FINAL_REPORT_21March2016.pdf

Format: PDF, 116 pages
Authority / posture: governor-appointed independent task-force report; not a judicial finding or regulatory order.

May support

  • task force's attribution of primary governmental failures;

  • MDEQ failures in LCR interpretation/enforcement;

  • MDHHS delay;

  • Flint Water Plant treatment failures;

  • EPA delayed enforcement;

  • environmental-justice framing;

  • 36 findings / 44 recommendations.

Cannot establish

  • criminal liability;

  • every agency/individual's legal responsibility.

Captureability: HIGH
Priority: P0


FL-S08 — Virginia Tech Flint Water Study, September 2015 citizen-science sampling

Source: Virginia Tech Flint Water Study / Marc Edwards team
Research surface: https://flintwaterstudy.org/
Relevant dataset/report pages: 2015 citizen-science sampling and public results.

Authority / posture: independent academic/research sampling, not official LCR compliance sampling.

Key reported results

  • 252 residential samples in the cited early dataset;

  • 42 samples (~16.7%) above 15 ppb;

  • first-draw 90th percentile around 25.2 ppb in that sampling set;

  • many samples above 5 ppb.

Critical qualification The VT sampling frame and protocol were not the same as Flint's formal LCR high-risk compliance sampling frame. Its 90th percentile should not be compared to the official LCR percentile as if the two sampled identical populations under identical rules.

Captureability: MEDIUM-HIGH
Hazards

  • blog/research-site pages can move;

  • acquire raw downloadable dataset and any contemporaneous methodology documentation;

  • preserve update/version history.

Priority: P0


FL-S09 — Virginia Tech 2016 resampling

Issuer: Virginia Tech Flint Water Study
Date: March 2016 resampling wave
Source surface: https://flintwaterstudy.org/

Authority / posture: independent longitudinal/resampling research.

May support

  • lead levels in participating homes were improving after treatment/source changes;

  • in the resampled cohort, first-draw 90th percentile declined relative to 2015 but remained above 15 ppb at that stage.

Cannot establish

  • whole-city LCR compliance from the research cohort;

  • every home's trajectory.

Captureability: MEDIUM-HIGH
Priority: P1


FL-S10 — Hanna-Attisha et al. peer-reviewed blood-lead study

Exact title: Elevated Blood Lead Levels in Children Associated With the Flint Drinking Water Crisis: A Spatial Analysis of Risk and Public Health Response
Authors: Mona Hanna-Attisha et al.
Journal: American Journal of Public Health
Publication: 2016
PubMed: https://pubmed.ncbi.nlm.nih.gov/26691115/
PMC full text: https://pmc.ncbi.nlm.nih.gov/articles/PMC4985856/

Authority / posture: peer-reviewed observational epidemiologic study.

Key finding Among children under 5 in the analyzed Flint area, elevated blood lead prevalence increased from about 2.4% before the water-source change to 4.9% after, with larger increases in neighborhoods with higher water-lead exposure.

Critical qualification Study population, geography, age range, time windows and analytic choices matter. This should not be conflated with CDC's later analysis.

Captureability: HIGH
Priority: P0


FL-S11 — CDC MMWR blood-lead analysis

Exact title: Blood Lead Levels Among Children Aged <6 Years — Flint, Michigan, 2013–2016
Issuer: CDC / MMWR
Exact current article: official CDC MMWR page for Flint 2013–2016 blood lead analysis.

Authority / posture: federal public-health epidemiologic analysis.

Key period results

  • before switch: ~3.1% elevated BLL;

  • early Flint River period: ~5.0%;

  • late Flint River period: ~3.9%;

  • after return to Detroit water: ~1.4%.

Adjusted models show period-specific differences and uncertainty.

Counterpedia value Numbers differ from Hanna-Attisha because:

  • age band differs;

  • exact periods differ;

  • dataset/geography/model differ.

This is a perfect "different denominators do not equal contradiction" specimen.

Captureability: HIGH
Priority: P0


FL-S12 — 2017 Concerned Pastors settlement agreement

Exact title: settlement agreement in Concerned Pastors for Social Action, et al. v. Khouri, et al.
Date: March 27, 2017
Exact PDF:
https://www.michigan.gov/flintwater/-/media/Project/Websites/flintwater/documents/2017/Settlement_Agreement_dated_032717.pdf

Format: PDF, 92 pages
Authority / posture: negotiated legal settlement.

Critical posture The agreement states it is entered without admission of any issue of fact or law.

May support

  • agreed service-line replacement, monitoring, filter, reporting and infrastructure obligations;

  • defendants' contractual obligations under settlement.

Cannot establish

  • admission of liability merely because replacement was agreed.

Captureability: HIGH
Priority: P1


FL-S13 — Michigan/EGLE current residential & business sampling page

Issuer: State of Michigan / EGLE Flint Water
Exact locator:
https://www.michigan.gov/flintwater/resources/testing-results/residential-and-business-sampling

Authority / posture: current state sampling-results/index surface.

Important explanatory value The page explains that the LCR 90th percentile is a system-level compliance statistic; the remaining 10% of sampled sites may have higher values, potentially much higher.

Current data The page links 2026 sampling results and historical files.

Captureability: HIGH
Hazard Some explanatory prose on the page appears stale relative to current linked data (for example, language referring to older monitoring periods). Capture page observation separately from child data artifacts.

Priority: P0


FL-S14 — Flint January–June 2026 individual sampling report

Exact title: Flint residential/business test results, January–June 2026, sorted by lead values
Issuer: State of Michigan / EGLE
Exact current child PDF: linked from FL-S13; current URL path includes /2026/testing-res-bus/Test-Results-Flint-Jan-Jun-2026-sorted-by-lead-values.pdf

Format: PDF, 4 pages
Authority / posture: current state-reported individual sample results.

Key current evidence The table contains individual lead results above Michigan's 12 ppb threshold, including values such as 16, 27, and 34 ppb, even while the distribution system remains in LCR compliance.

Data semantics The report notes:

  • values >12 ppb highlighted;

  • "Not Detected" converted to numeric 0 for sorting convenience.

Counterpedia value This is the clearest current example that: system 90th percentile compliant does not mean every sampled premise below the action level.

Captureability: HIGH
Priority: P0


FL-S15 — Michigan February 5, 2026 compliance update

Exact title: Flint water enters 10th year of lead testing compliance
Issuer: State of Michigan / EGLE
Date: February 5, 2026
Current state-news locator: Michigan Flint Water news release.

Authority / posture: current state regulatory/status announcement.

Key current facts

  • July–December 2025 90th percentile: 6 ppb;

  • Flint has remained below applicable action levels since July 2016;

  • Michigan's tighter 12 ppb action level became effective in 2025; federal LCR reference remains 15 ppb in the relevant comparison;

  • latest compliance calculation used 61 known lead-service-line properties;

  • nearly 11,000 service lines under the settlement had been replaced by July 2025;

  • state/city reporting indicated roughly 98% of residential lead service lines replaced, while some lines remained, including opt-outs/new discoveries.

Critical qualification A 6 ppb 90th percentile is not an assertion that every tap contains ≤6 ppb or that lead exposure risk is zero.

Captureability: HIGH
Priority: P0


FL-S16 — EPA lead action-level / health-goal semantics

Issuer: U.S. EPA
Current official lead/copper rule and lead-health pages

Key regulatory semantics

  • lead Maximum Contaminant Level Goal (MCLG): zero;

  • no known safe level of lead exposure;

  • LCR action level is a treatment-technique / corrosion-control compliance trigger, not an individual-home health standard.

Authority / posture: federal regulatory/health-policy explanation.

Counterpedia value This source prevents the very common laundering:

"below 15 ppb" → "safe amount of lead."

Captureability: HIGH
Priority: P0


FL-S17 — EPA filter challenge / filter effectiveness studies

Issuer: U.S. EPA
Date: 2016 Flint filter challenge reports
Official EPA Flint source surface: Flint drinking-water documents

Authority / posture: federal intervention/testing evidence.

May support

  • appropriately installed/distributed certified filters were effective at substantially reducing/removing lead under tested high-risk Flint conditions.

Cannot support

  • unfiltered water was safe;

  • filters solve infrastructure corrosion permanently.

Captureability: HIGH
Priority: P2


FL-S18 — Wikipedia comparison surface — OPTIONAL

Title: Flint water crisis
Role: secondary comparison surface only
Capture rule: pin exact oldid if used.
Authority / posture: collaborative secondary narrative; no automatic admission.
Priority: OPTIONAL


C. RECORD-CONTRADICTION / QUALIFICATION MAP

C-1 — Official 2015 LCR results vs Virginia Tech 2015 results

EPA OIG records official Flint 90th-percentile results of:

  • 6 ppb in the first 2015 monitoring round;

  • 11 ppb in the second.

Both were below the then-federal 15 ppb LCR action level.

Virginia Tech's separate 2015 residential sampling set reported:

  • a first-draw 90th percentile around 25.2 ppb;

  • 42 of 252 samples above 15 ppb.

A normal article can make those numbers look mutually exclusive.

They are not directly equivalent because:

  • the samples were not the same homes;

  • the site-selection logic differed;

  • official LCR sampling was supposed to target specific high-risk tiers;

  • EPA/OIG later identified problems in Flint's high-risk-site inventory;

  • EPA later warned that flushing, aerator removal and bottle configuration can bias samples lower.

Counterpedia rule Every lead number needs: population

  • site-selection rule

  • sampling protocol

  • statistic

  • date.

Do not resolve the apparent contradiction by picking whichever number fits the narrative.


C-2 — "Below the action level" vs "safe"

EPA's LCR action level is not a declaration that water with less lead is biologically safe.

EPA's health goal for lead is zero and the agency states there is no known safe level of lead exposure.

Therefore:

system 90th percentile < action level means: the system meets the relevant corrosion-control compliance trigger for that monitoring period

not: every household has a safe lead level.

This is likely Flint's strongest denominator/semantic reveal.


C-3 — Current system compliance vs current household outliers

Michigan's current status record says:

The January–June 2026 individual result file still contains samples such as:

  • 16 ppb;

  • 27 ppb;

  • 34 ppb.

Those are not contradictory.

A 90th percentile statistic allows some values above the percentile and the action level.

Counterpedia can make the hierarchy explicit:

city/system monitoring period90th percentile = 6 ppb

while:

individual premise sample34 ppb

Both can be true.


C-4 — "Flint's water is safe now" vs temporal and spatial scope

By 2025–2026:

  • EPA terminated the 2016 emergency order;

  • Michigan says Flint has maintained LCR compliance for years;

  • corrosion-control and infrastructure conditions are dramatically different from 2014–2015;

  • most residential lead service lines have been replaced.

But:

  • some service lines remain;

  • individual sample outliers still exist;

  • the action level is not a zero-risk health threshold.

So a current article should refuse the unscoped binary: safe / unsafe.

Better claim objects are:

  • system complied with LCR during monitoring period X;

  • emergency order terminated on date Y;

  • individual premise P measured Z ppb on date D;

  • service-line replacement status at date D;

  • EPA health goal for lead remains zero.


C-5 — Blood lead prevalence values differ across studies without necessarily conflicting

Hanna-Attisha et al.:

  • children <5;

  • one spatial/time design;

  • about 2.4% → 4.9%.

CDC MMWR:

  • children <6;

  • different period definitions/dataset/model;

  • 3.1% before;

  • 5.0% early river period;

  • 3.9% late river period;

  • 1.4% after switch back.

These values answer related but non-identical questions.

Counterpedia rule Never create a synthesized "the childhood lead rate doubled from 3.1 to 4.9%" sentence by mixing denominators across studies.


EPA OIG and the Flint Water Advisory Task Force strongly criticize treatment, regulatory interpretation and government response.

The 2017 Concerned Pastors settlement imposed extensive service-line and monitoring obligations but says it was entered without admission of fact or law.

Therefore:

  • technical/administrative causation findings;

  • settlement obligations;

  • admission;

  • legal liability

must remain distinct.


C-7 — OIG finding vs EPA operational closure

EPA could certify corrective recommendations as completed.

EPA OIG later found three actions still did not fully address the original oversight deficiencies.

This is not necessarily "EPA contradicting itself" in one voice.

It is: management closure vs independent inspector-general sufficiency review.


C-8 — Emergency order active vs emergency order terminated

January 21, 2016:

  • EPA finds state/city response inadequate and imposes emergency measures.

May 2025:

  • EPA announces termination after required actions/remediation milestones were completed.

Both are true at their dates.

The sentence:

"EPA says Flint is under an emergency drinking-water order"

is therefore edition/time-sensitive.


C-9 — Current EPA metadata conflict on termination date

EPA's current Flint documents page appears to label the termination document as September 1, 2021.

But the linked PDF path/content and EPA's May 19, 2025 public announcement indicate the relevant termination action is later.

This is an outstanding Counterpedia source-integrity specimen:

webpage metadata observation ≠ necessarily document's internal dateevent date.

Do not "clean up" the source silently.


C-10 — Service-line replacement statistics need denominator/date

"98% replaced" is not equivalent to:

  • all service lines;

  • all residential premises;

  • all galvanized plumbing;

  • all remaining lead-bearing fixtures.

Statements need:

  • denominator;

  • cutoff date;

  • scope;

  • opt-outs/new discoveries.


D. FIVE-CLAIM DEMONSTRATION MATRIX — PROVISIONAL ONLY

C1 — Observable / chronology fact

Provisional wording

Flint switched its drinking-water source to the Flint River in April 2014 without implementing continuous corrosion-control treatment, and returned to Detroit/Great Lakes Water Authority-supplied water in October 2015.

Candidate sources

Can establish

  • source-change chronology;

  • lack of continuous CCT after switch;

  • later return to Detroit-supplied water.

Cannot establish

  • every downstream health outcome;

  • legal liability.

Likely posture

  • strong institutional chronology/technical fact.

Qualification Use exact source-switch and return dates from governed capture before final composition.


C2 — Measurement / denominator claim

Provisional wording

Flint's official 2015 LCR monitoring reported 90th-percentile lead values of 6 ppb and 11 ppb, while an independent Virginia Tech residential sampling effort reported a substantially higher first-draw 90th percentile of about 25.2 ppb; the two results came from different sampling frames and protocols and should not be treated as directly interchangeable citywide measurements.

Candidate sources

Can establish

  • exact reported statistics;

  • sampling-frame/protocol differences;

  • official site-inventory/sampling weaknesses.

Cannot establish

  • that one number is the one "true Flint lead level";

  • intentional fraud merely from discrepancy.

Likely posture

  • multi-source measurement comparison.

Refusal branch

"The city tested 11 ppb while scientists tested 25.2 ppb, proving the city falsified the data."

The discrepancy alone does not establish intentional falsification.

Strengthening Would require exact sample-level matching, chain-of-custody/protocol evidence, and proof of knowing manipulation if fraud is alleged.


C3 — Action / regulatory-response claim

Provisional wording

State regulators did not require Flint to use continuous corrosion-control treatment after the 2014 source switch; EPA OIG later found state and federal management/oversight weaknesses delayed an effective response, and EPA issued a Safe Drinking Water Act emergency order on January 21, 2016.

Candidate sources

Can establish

  • regulator/treatment chronology;

  • OIG/task-force findings;

  • emergency-order action.

Cannot establish

  • criminal intent;

  • one legal-liability allocation for all entities.

Likely posture

  • attributed oversight/administrative finding + formal regulatory action.


C4 — Carefully bounded causal / public-health claim

Provisional wording

The institutional and epidemiologic record supports a causal chain in which the untreated Flint River source increased corrosion of lead-bearing plumbing, increasing lead exposure risk; peer-reviewed and CDC analyses found higher elevated-blood-lead prevalence among young children during the Flint River period, with estimates varying by study population and method.

Candidate sources

Can establish

  • corrosion-control / infrastructure pathway;

  • period-associated elevated BLL increases;

  • study-specific estimates.

Cannot establish

  • exact individual dose from system-level water data;

  • that every child with an elevated BLL was exposed solely through Flint drinking water;

  • a single universal citywide increase percentage.

Likely posture

  • bounded causal synthesis with study-specific epidemiologic qualification.

Refusal branch Avoid:

"The water crisis doubled every Flint child's blood lead level."

That is not what either major study measured.


C5 — Contested / commonly compressed proposition

Popular formulation

"Flint's water is safe now."

What the current record supports

  • EPA terminated the 2016 emergency order in 2025;

  • Flint has maintained system-level LCR compliance for years;

  • 2025 H2 90th percentile was 6 ppb;

  • extensive service-line replacement has occurred.

What the current record does not establish at that scope

  • every tap has zero lead;

  • every premise is below 12 or 15 ppb;

  • all lead service lines are gone;

  • the LCR action level is a health-based safe threshold;

  • no person has residual exposure risk.

Current counterexample 2026 individual sampling includes some reported values above Michigan's 12 ppb action level.

Likely posture

  • scope reduction / refusal of unqualified "safe."

Stronger admissible wording

Flint's distribution system is currently meeting the applicable Lead and Copper Rule system-level compliance metric, but that metric does not mean every premise has zero lead or that every individual sample is below the action level.

Reopening condition "Every Flint tap is below X" requires:

  • defined X and health/regulatory meaning;

  • exhaustive or statistically valid premise-level sampling at the relevant date;

  • defined sampling protocol;

  • treatment of service lines/fixtures and temporal variability.

For "zero lead risk," an action-level compliance statistic cannot supply the required evidence because EPA's health goal for lead is zero and exposure can occur at premise-specific plumbing.


E. BEST "WHY NOT?" SPECIMEN

"Flint's water passed lead testing in 2015, so officials had no evidence of a serious lead problem."

Why not

The official LCR 90th percentiles were indeed below the then-federal action level.

But the record also shows:

  • incomplete/incorrect high-risk-site inventory;

  • sampling-protocol problems capable of lowering observed lead;

  • no continuous corrosion-control treatment after the source switch;

  • independent household sampling with much higher values;

  • later public-health evidence;

  • later institutional findings that the response was seriously delayed.

So: official compliance statistic below trigger does not imply absence of contrary evidence or absence of risk.

Reopening condition

To promote the stronger historical claim, the evidence would have to establish that:

  • the official sample frame complied with required high-risk site selection;

  • protocol did not systematically bias results;

  • relevant independent/higher results were unavailable or invalid;

  • corrosion control and infrastructure conditions did not create an unmeasured risk.

The located institutional record points the other way.


F. SECOND "WHY NOT?" — CURRENT STATUS

"Flint's water is still poisoned."

This is also too broad if presented as a current system-level regulatory claim.

Current state/EPA records support:

  • prolonged LCR compliance;

  • emergency-order termination;

  • large-scale service-line replacement.

But a blanket opposite sentence—"there is no lead problem anywhere"—is also too broad because individual samples can remain elevated.

Counterpedia reveal The page need not choose a slogan.

It can say exactly:

  • system compliance at date X;

  • 90th percentile Y;

  • individual sample Z;

  • service-line inventory status;

  • current regulatory action;

  • health-goal semantics.


G. SOURCE-PAGE / REVERSE-WIKIPEDIA VALUE

1. FL-S05EPA 2016 sampling memorandum

Why ideal

A deceptively small technical memo explains why:

  • flushing;

  • aerator removal;

  • bottle geometry;

  • stagnation

can change a lead result.

Reverse traversal:

sampling-method ruleofficial Flint sampling criticismspecific 2015 resultmeasurement claimclaim qualification/refusal

This is an excellent "method is part of the fact" Source Page.


2. FL-S01EPA OIG 2018 report

Why ideal

It connects:

  • source-water change;

  • corrosion control;

  • site inventory;

  • official 2015 6/11 ppb results;

  • MDEQ actions;

  • EPA oversight;

  • delayed response;

  • recommendations.

The Source Page can show what is:

  • OIG finding;

  • quoted EPA/state position;

  • underlying monitoring datum;

  • recommendation.


3. FL-S14 — 2026 individual sample-results artifact

Why ideal

A current data file can reverse-traverse from one premise's 34 ppb result into:

  • sampling date/protocol;

  • current Michigan action level;

  • citywide 90th-percentile computation;

  • why that result can coexist with system compliance;

  • service-line/remediation status.

This is a perfect quantitative Reverse-Wikipedia object.


4. FL-S10 + FL-S11 — Hanna-Attisha / CDC epidemiology pair

Why ideal

A paired Source Page can expose:

  • age inclusion;

  • time windows;

  • geographic units;

  • elevated-BLL definition;

  • raw percentages;

  • adjusted odds;

  • why the values differ.

The UI can prevent users from synthesizing incompatible denominators into a false number.


H. 30–60 SECOND DEMO MOMENT

Open Flint and show four numbers:

2015 official LCR

90th percentile: 11 ppb

2015 Virginia Tech independent sampling

first-draw 90th percentile: ~25.2 ppb

2025 H2 current compliance

90th percentile: 6 ppb

2026 individual premise

one reported sample: 34 ppb

Ask:

Which one is Flint's lead level?

Counterpedia answers:

None of those is "the lead level." They are different measurements with different populations, protocols, dates and statistical meanings.

Then expose the rules:

11 ppb → formal 2015 LCR set → official site-selection/protocol problems later identified

25.2 ppb → independent citizen-science cohort → different sampling frame

6 ppb → current city/system 90th percentile → regulatory compliance statistic

34 ppb → individual 2026 premise sample → allowed to coexist statistically with a 6 ppb 90th percentile

Then click:

"So is Flint's water safe now?"

Counterpedia:

The system currently meets the LCR compliance metric. That metric is not a statement that every tap has zero lead or that every individual sample is below the action level.

That is the demo.


I. CAPTURE PRIORITY

P0 — essential

  1. FL-S01 EPA OIG 2018 exact PDF

  2. FL-S03 EPA Jan. 21, 2016 emergency order

  3. FL-S04 emergency-order termination:

    • current page observation

    • linked PDF bytes

    • May 19, 2025 EPA release

  4. FL-S05 EPA Feb. 29, 2016 sampling memo

  5. FL-S07 Flint Water Advisory Task Force final report

  6. FL-S08 Virginia Tech 2015 methodology/results + raw dataset

  7. FL-S10 Hanna-Attisha peer-reviewed article

  8. FL-S11 CDC MMWR analysis

  9. FL-S13 current Michigan sampling/index page

  10. FL-S14 Jan–Jun 2026 individual sample report

  11. FL-S15 Feb. 5, 2026 compliance update

  12. FL-S16 EPA action-level/MCLG semantics

P1 — important corroboration / contradiction

  1. FL-S02 EPA OIG 2022 follow-up

  2. FL-S06 EPA MDEQ program review

  3. FL-S09 VT 2016 resampling

  4. FL-S12 2017 settlement agreement

P2

  1. FL-S17 filter challenge/effectiveness studies

  2. underlying sequential-sampling data / premise-level plumbing studies if a more granular demo is pursued

OPTIONAL

  1. FL-S18 Wikipedia comparison edition pinned by oldid


J. INTEGRITY FLAGS

1. Action level is not a "safe lead" threshold

Mandatory qualification on every use of 12/15 ppb.


2. 90th percentile is not the maximum

Current individual values above a citywide 90th-percentile result are mathematically compatible.


3. Official and VT 2015 percentiles are not directly interchangeable

Do not compare as if identical site populations/protocols.


4. Official site inventory was itself defective

EPA OIG found Flint lacked the required accurate high-risk service-line inventory.


5. Sampling method can bias results

Aerator removal, pre-stagnation flushing, bottle configuration and site selection matter.


6. "Below action level" in 2015 does not erase later institutional findings

The compliance statistic and the later criticism refer to different aspects of the monitoring/treatment system.


7. Blood-lead studies use different denominators

Hanna-Attisha and CDC percentages must retain:

  • age range;

  • dates;

  • geography;

  • elevated-BLL definition;

  • model.


8. Association vs individual causation

Population-level increase does not prove source/dose for every child.


9. OIG posture is OIG posture

Do not label OIG conclusions as a unanimous/final EPA institutional determination.


10. Task Force posture is independent advisory/investigative

Not a court judgment.


11. Settlement contains no admission

Do not infer liability admission from replacement obligations.


12. EPA termination metadata conflict must remain visible

The current webpage's September 1, 2021 label appears inconsistent with the linked 2025 document path/content and May 2025 EPA announcement.

Capture all layers before adjudicating.


13. Current state page contains stale historical prose

Do not assume current-page prose and current linked datasets share the same update date.


14. "Not Detected" → 0 is a presentation transformation

The 2026 sorted data file explicitly converts ND to numeric zero for sorting. Zero in that derived presentation is not necessarily an instrument measurement of exactly 0 ppb.


15. Michigan 12 ppb and federal 15 ppb need date/jurisdiction labels

Do not display one unlabeled "action level."


16. Service-line replacement claims need denominator

"98%" needs scope/date and remaining opt-outs/new discoveries.


17. Emergency-order termination does not erase historical emergency status

Historical page edition should still show the order as active in the period it was active.


18. Current compliance is time-sensitive

Recapture current EGLE/EPA pages and child datasets before publication.


K. FLINT CAPTURE DISPATCH SHAPE

For FLINT-CAP1, acquire:

  1. FL-S01 exact OIG 2018 PDF

  2. FL-S05 EPA sampling memo

  3. FL-S03 2016 emergency order

  4. FL-S04:

    • current EPA index HTML

    • linked termination PDF

    • May 19, 2025 announcement

    • preserve metadata disagreement

  5. FL-S07 Task Force PDF

  6. FL-S06 EPA MDEQ program review

  7. FL-S08 exact VT 2015 methodology page + raw dataset + contemporaneous result pages

  8. FL-S09 VT 2016 resampling

  9. FL-S10 peer-reviewed article

  10. FL-S11 CDC MMWR

  11. FL-S12 settlement PDF

  12. FL-S13 current Michigan page

  13. FL-S14 exact Jan–Jun 2026 PDF

  14. FL-S15 current compliance announcement

  15. FL-S16 EPA action-level / MCLG sources

  16. FL-S02 OIG follow-up

  17. FL-S17 filter-study artifacts if needed

CAP1 output remains intentionally boring:

  • requested locator

  • final locator

  • HTTP result

  • media type

  • exact bytes/digest

  • capture observation

  • source-owned dates

  • monitoring period

  • sampling protocol

  • sample population

  • statistic/denominator

  • regulatory threshold + jurisdiction/date

  • source-version/metadata conflicts

  • extraction/table anchorability

  • captured / refused / failed

  • no final standing

  • no article prose


L. PROVISIONAL FLINT THESIS FOR REVIEW

Flint should not be demoed as:

"The water had X ppb of lead."

That sentence is almost meaningless without:

  • where;

  • when;

  • which tap;

  • first draw or sequential;

  • stagnation period;

  • sampling cohort;

  • percentile or individual value;

  • regulatory threshold;

  • health interpretation.

The stronger Counterpedia thesis is:

A city can be in regulatory compliance while some individual homes still have higher lead values, and a city can report a compliance percentile below an action level while the sampling program itself is later found defective.

The most revealing page moment is that 11 ppb, 25.2 ppb, 6 ppb, and 34 ppb can all be faithfully sourced without contradiction.

Flint therefore tests:

  • temporal truth;

  • statistical/denominator discipline;

  • sampling-method provenance;

  • regulatory semantics vs health semantics;

  • system-level vs household-level status;

  • research vs compliance sampling;

  • observational epidemiology vs individual causation;

  • administrative closure vs inspector-general review;

  • settlement vs admission;

  • source-page metadata drift;

  • and the need for every quantitative claim to carry its method and population with it.

The core invariant is:

A number without its denominator, protocol, time, and authority posture is not yet a governed fact.