Methodology Archive · /judicial/#actors

Every claim, every query, every row-count.

Below: the live SQL backing every load-bearing figure on /judicial/#actors, the methodological caveats that shape what the data can and cannot say, and the technical results moved here when they failed the page's "five-second test." Project: ibfmjtwahkwqzcmeyqii. Read-only. SELECT only. The Supabase MCP tooling makes the same queries directly runnable from any modern AI agent.

← back to /judicial/#actors
00 — Frozen Window

What "frozen window" means in practice

The MCRO court-records corpus underlying this page was downloaded between April 26 and April 30, 2024. (See /the-download/.) Since that download, only Matt Guertin's case (27-CR-23-1886) has been kept current. Some isolated additional children/case files exist; they are exceptions, not a refresh.

This means:

This caveat applies to every claim that follows.

01 — Glossary

The terms used on the main page, with the technical detail underneath each one

Judicial officer (JO)
Judges, referees, magistrates — anyone authorized to sign court orders. In the database, normalized via parents_case_events_judicial_officers.judicial_officer_json__judicial_officer_norm.
Event
Any officer-attributed action on a docket: signing an order, presiding over a hearing, filing a finding. Stored across parents_case_events (events) and parents_case_events_judicial_officers (officer attribution). The two are dependent; counts are not additive across them.
Roster
Minnesota Fourth Judicial District public list of currently-serving judges and referees, captured November 28, 2025 from mncourts.gov: 90 PDF biographies. Stored on this site at /provenance/data/06-fourth-district-judicial-roster/.
Cluster
A custom (cluster_id, cluster_name) pair stored on parents.cluster_id / parents.cluster_name that ties together cases belonging to the same defendant when the official records are filed under disjointed name variants (suffixes added or dropped, middle initials varied, slight spelling drift). Without the cluster layer, a defendant carrying 16 separate case numbers can appear as 3 or 4 unrelated people in the docket. The 77-cluster forensic set referenced throughout this page is documented at /77/.
Signature edit
A modification to a PDF made after the digital signature was applied. Cryptographically detectable. Recorded in children_signatures_full_report.edits_after_sig_flag. Includes substantive document edits AND watermark/timestamp operations — the page does not over-claim "tampering" without further inspection.
Frozen window
The MCRO download is fixed at April 29–30, 2024. Some comparisons are not available because the data has not been refreshed since (see §00).
edits_after_sig_flag
Boolean column on children_signatures_full_report. true = post-signature edits detected. false = no post-signature edits. Headline finding: this flag is true for 1,079 of 1,080 non-watermark per-officer signatures across the 77-cluster forensic set (99.91%).
sig_crypto_valid_flag
Boolean column on children_signatures_full_report. Schema convention: NULL = crypto-validation absent / failed (not literally false). Page treats NULL as crypto-fail; the methodological caveat is that direct PDF inspection is required for definitive crypto-fail confirmation.
02 — Roster Math

90 / 247 / 168 — how the surplus is computed

Claim · Published roster size
The Hennepin Fourth Judicial District publishes 90 active-bench biographies on mncourts.gov.
Holds
Roster directory listing reads "1-90 of 90 results" on the November 28, 2025 capture. PDF biographies preserved in /provenance/data/06-fourth-district-judicial-roster/pdf/ (count = 90).
Claim · Distinct human officers in the database
The corpus contains 247 distinct real-human judicial officers after normalization and removal of phantom-routing IDs (Presiding Judge, Scheduling Judge, Sip Conv. Hennepin).
Downgraded from 274
-- Distinct normalized JOs in the JO spine, excluding 3 known routing placeholders SELECT COUNT(DISTINCT judicial_officer_json__judicial_officer_norm) FROM parents_case_events_judicial_officers WHERE judicial_officer_json__judicial_officer_norm NOT IN ('Presiding Judge', 'Scheduling Judge', 'Sip Conv. Hennepin');
Result: 247. Page-1 prior-art figure of 274 came from a denominator that included middle-initial-fronted normalization variants (Regis 139, H. Albrecht 134, Ekstrum 7, Hopper 6, Mckinsey 2). Strict denominator = 247.
Claim · 76 of 90 published names match the database
76 names on the published roster are present in the JO spine; 168 distinct officers in the database have no published roster entry.
Holds (strict match)
76 strict-matched / 79 fuzzy-matched (Levenshtein ≤4 across normalized variants). Page uses 76 strict. Surplus under the strict match: 247 − 76 published-and-matched = 168 + 3 phantoms = 171 raw; 168 after phantom removal.
Claim · Borer is the #1 surplus officer
George Borer's appearance count in the JO spine: 1,085 across 434 distinct cases. Roughly 30% of those events are competency-pipeline events.
Downgraded from 5,830
-- Borer per-event count in the JO spine SELECT COUNT(*) AS total_events, COUNT(DISTINCT case_uid) AS distinct_cases FROM parents_case_events_judicial_officers WHERE judicial_officer_json__judicial_officer_norm = 'George Borer';
total_events: 1,085 · distinct_cases: 434. The 5,830 figure cited in the prior-art Judicial Officer Representation Analysis PDF was an aggregated count across multiple downstream tables (source_hearings + parents_case_events + parents_hearings_previous), each of which carries its own dependent denominator. Strict JO-spine count is 1,085 (rank #8 corpus-wide; rank #1 in the surplus list, since the seven officers above him are all on the published roster).
Caveat · "Top 30 surplus competency referees" framing on prior page
The earlier page claimed "10 of the top 30 surplus officers are competency-pipeline referees, carrying 8,095 combined appearances." The 8,095 figure is not reproducible against the JO spine; the actual combined volume of the 10 highest-volume competency-heavy surplus officers is ~1,843 events (or 6,097 across the full top 30 surplus). The page no longer cites the 8,095 figure.
Dropped from main page
03 — Clique Density

21-of-21 edges · internal weight 4,263 vs. comparator 227

Claim · Fully connected sub-graph at ≥5 shared cases
The seven-officer ring {Klein, Mercurio, Browne, Janzen, Koch, Skibbie, Lamas} forms 21 of 21 possible pairs in the co-appearance graph at the threshold of ≥5 shared cases. Edge weights range from min 40 to max 477.
Holds (exact)
-- Build the ring's pairwise co-appearance edge list at the ≥5-shared-cases threshold WITH ring(officer) AS (VALUES ('Julia Dayton Klein'),('Danielle Mercurio'),('Michael Browne'), ('Lisa Janzen'),('William Koch'),('Lori Skibbie'),('Carolina Lamas')), shared AS ( SELECT a.judicial_officer_json__judicial_officer_norm AS jo_a, b.judicial_officer_json__judicial_officer_norm AS jo_b, COUNT(DISTINCT a.case_uid) AS shared_cases FROM parents_case_events_judicial_officers a JOIN parents_case_events_judicial_officers b ON a.case_uid = b.case_uid AND a.judicial_officer_json__judicial_officer_norm < b.judicial_officer_json__judicial_officer_norm JOIN ring ra ON ra.officer = a.judicial_officer_json__judicial_officer_norm JOIN ring rb ON rb.officer = b.judicial_officer_json__judicial_officer_norm GROUP BY 1, 2 ) SELECT COUNT(*) FILTER (WHERE shared_cases >= 5) AS edges_at_threshold, SUM(shared_cases) FILTER (WHERE shared_cases >= 5) AS internal_weight, MIN(shared_cases) FILTER (WHERE shared_cases >= 5) AS min_edge, MAX(shared_cases) AS max_edge FROM shared;
edges_at_threshold: 21 · internal_weight: 4,263 · min_edge: 40 · max_edge: 477.
Claim · Matched comparator carries internal weight 227
The matched comparator group {Hoyos, Kappelhoff, Chou, Burns, Hudleston, Askalani, Brandt} — selected at the same caseload tier — carries an internal weight of 227 at the same ≥5-shared-cases threshold. Ratio: 18.78×. 18 of the 21 comparator pairs cross the threshold (the ring is at 21 of 21).
Holds (exact)
Same query as above with the comparator officer set substituted. Result: 18 edges crossing ≥5 / total weight 227. Ring weight 4,263 / comparator weight 227 = 18.78×.
Claim · 99.95th-percentile null test (2,000 random tuples)
The Judicial Officer Representation Analysis report attests that 2,000 random 7-officer tuples drawn from officers with ≥100 case appearances produced a null distribution with mean ~783 and SD ~630, placing the ring's 4,263 beyond the 99.95th percentile.
Spot-checked, attested
Five spot-check 7-tuples re-run on the live database yielded internal weights of 292 / 332 / 538 / 634 / 1,535. The clique's 4,263 sits well above all five. The full 2,000-trial null vector is not preserved in the source raw_output; the page treats this as document-attested with five spot-check confirmations.
Claim · Top-10 case-sharing pairs all involve ring members; Klein is the hub
The corpus-wide top-10 officer pairs by shared case count are dominated by the ring: Browne–Klein (477), Mercurio–Klein (437), Browne–Mercurio (348), Koch–Klein (299), Janzen–Klein (288), Skibbie–Klein (269), Browne–Borer (249), Borer–Klein (244), Koch–Mercurio (235), Browne–Skibbie (235). 10 of 10 involve a ring member. Klein appears in 6 of 10.
Holds (exact)
04 — Browne Apex

125 cases / 0 of 601 hearings · the source_hearings vs. parents_hearings_previous note

Claim · 125 distinct cases, 177 signatures, 80% in 2023
Michael Browne carries cryptographic per-officer signatures on 125 distinct cases across 177 individual signature rows. 79.7% (141 of 177) of those signatures fall in calendar year 2023.
Holds (exact)
SELECT COUNT(DISTINCT case_uid) AS cases, COUNT(*) AS sigs, ROUND(100.0 * COUNT(*) FILTER (WHERE EXTRACT(YEAR FROM sign_time) = 2023) / COUNT(*), 1) AS pct_2023 FROM children_signatures_full_report WHERE signer_name ILIKE '%Browne%';
cases: 125 · sigs: 177 · pct_2023: 79.7
Claim · 0 of 601 source_hearings; Mercurio 251 / Klein 195 / Borer 146
Across the 125 cases Browne signed, the court's source_hearings index records 601 hearings. Browne held zero of them. Mercurio held 251, Klein 195, Borer 146 — 592 of 601.
Holds (source_hearings denominator)
Required disclosure · parents_hearings_previous discrepancy
The parallel parents_hearings_previous table records ~130 Browne hearings across 2,330 total hearings on the same 125 cases (Mercurio 246, Klein 220, Borer 160). Browne is the smallest hearing-presider in BOTH tables, but the absolute-zero figure is specific to source_hearings.
Footnoted on main page
The page leads with the source_hearings 0/601 figure and footnotes the parents_hearings_previous count. The structural finding (Browne is the smallest hearing-presider on his own signed cases) holds in BOTH tables; only the absolute-zero figure is table-specific.
Claim · Mercurio 100% / Klein 96.8% / Borer 78.4% on the 125
Of Browne's 125 signed cases, Mercurio appears on every one (125/125), Klein on 121 (96.8%), Borer on 98 (78.4%).
Holds (exact)
Claim · Mean Mercurio-hearing-to-Browne-signature latency = -70.7 days
Across the 125 cases, the mean gap between Mercurio's first hearing on the case and Browne's signature on the order is -70.7 days. The hearings precede the signature by roughly two months.
Holds (matches -69.5 within rounding)
Claim · 31.2% ghost-signer rate (39 of 125)
On 31.2% of his 125 cases — 39 of them — Browne is a ghost-signer: his name is on the cryptographic signature but absent from the case's docket-officer array.
Downgraded from 62%
The 62% figure cited in prior reports used a 78-case sub-population denominator. Strict 125-case denominator yields 39/125 = 31.2%.
Cut from main page · "115× anomalous per-event signing rate"
The earlier page cited Browne's per-event signing rate as ~115× the typical Hennepin presider rate. Best derivable ratio against the live database is ~43×, not 115×. The 2.3 sigs/event figure is denominator-specific and not cleanly reconstructible across the multiple sig/event tables.
Dropped from main page
Page now leads with the structural fact (177 sigs / 0 source_hearings / 125 cases / 80% in 2023) instead of the multiplier.
05 — The Edits-After-Signature Wall (Per-Signer Detail)

100% across 8 named signers, 463 orders · full breakdown

Claim · 100% per-signer edits-after-sig across the 77-cluster set
Across the 77-cluster forensic decomposition, every per-officer cryptographic signature carries the edits-after-signature flag. Per-signer breakdown:
Holds (exact)
-- 77-cluster MCRO seed scope, per-signer edits-after-sig rate WITH mcro_seed AS ( SELECT DISTINCT p.cluster_id FROM parents p JOIN parents_case_set pcs ON p.case_uid = pcs.case_uid WHERE TRIM(BOTH '"' FROM pcs.case_set_json) = 'MCRO' AND p.cluster_id IS NOT NULL ), expanded AS ( SELECT p.case_uid FROM parents p WHERE p.cluster_id IN (SELECT cluster_id FROM mcro_seed) ) SELECT signer_name, COUNT(*) AS total_sigs, COUNT(*) FILTER (WHERE edits_after_sig_flag = TRUE) AS edited_sigs, COUNT(DISTINCT case_uid) AS cases, COUNT(DISTINCT e.cluster_id) AS clusters FROM children_signatures_full_report sig JOIN expanded e ON sig.case_uid = e.case_uid WHERE signer_name IN ('Michael Browne', 'Julia Dayton Klein', 'George Borer', 'Danielle Mercurio', 'Lisa Janzen', 'Hilary Caligiuri', 'Carolina Lamas', 'William Koch') GROUP BY signer_name ORDER BY total_sigs DESC;
SignerSigsEdited%CasesClusters
Michael Browne150150100%9845
Julia Dayton Klein144144100%8340
George Borer4444100%4222
Danielle Mercurio3434100%3112
Lisa Janzen3131100%2814
Hilary Caligiuri2626100%136
Carolina Lamas2424100%148
William Koch1010100%75
Total: 463 orders / 463 edited / 100%. Per-signer rate is 100% with no exceptions across all eight named signers.
Claim · 1,079 of 1,080 (99.91%) corpus-wide non-watermark exception
Across the entire 77-cluster set, 1,079 of 1,080 non-watermark per-officer signatures carry the edits-after-signature flag. The single exception is on Guertin's docket: a Minnesota Court of Appeals order signed by Chief Judge Frisch on the July 8, 2025 mandamus denial in case A25-0882. That order's edits_after_sig_flag = FALSE; its sig_crypto_valid_flag = NULL (treated as crypto-fail under the project's convention).
Holds
Aggregate framing as "99.98% (4,861 of 4,862)" cited in earlier reports comes from a denominator that mixed in the 3,780 MCRO Watermark application-layer signatures (which are false-by-definition). Strict per-officer (non-watermark) denominator: 1,080. Edited: 1,079. Rate: 99.91%.
Cut from main page · NULL crypto-valid rate "~38%"
Earlier reports cited a NULL crypto-valid rate of "~38% (409 of 1,080)" across per-officer rows. Re-derivation against the live database yields 17.42% (100 of 574 per-officer rows in the strict-per-officer subset). The methodological caveat itself (NULL = crypto-validation absent; PDF-inspect to confirm) remains correct.
Dropped from main page
06 — Industrial Signing

Borer 47-day, May 2 80-sig peak, the cluster patterns the page doesn't have room for

Claim · Borer mass-finding days
Borer single-day cluster counts of incompetency-related findings (Found Incompetent + Finding of Incompetency and Order combined): May 30, 2023 = 47; June 1, 2023 = 40; March 21, 2023 = 27; February 14, 2023 = 23.
Holds (exact); event-name combined
SELECT pce.event_json__date, COUNT(*) AS n FROM parents_case_events pce JOIN parents_case_events_judicial_officers jo ON jo.case_uid = pce.case_uid AND jo.event_index = pce.event_index WHERE jo.judicial_officer_json__judicial_officer_norm = 'George Borer' AND (pce.event_name ILIKE '%Found Incompetent%' OR pce.event_name_pretty ILIKE '%Finding of Incompetency and Order%') GROUP BY 1 ORDER BY n DESC LIMIT 10;
Claim · Cluster 76 (Makis Lane) 7+7 R20.01 issuance
On July 18, 2023, Browne signed 7 Rule-20.01 orders for the same defendant on the same 7 case IDs (cluster 76, Makis Lane). On October 10, 2023, Skibbie signed the same 7 R20.01 orders on the same 7 case IDs. Three months apart. Same defendant. Same paperwork. Different signing officer.
Holds (exact)
7 case IDs across cluster 76: 27-CR-19-11566, 19-12130, 20-20851, 20-9036, 21-1230, 21-13752, 22-10055.
Claim · Cluster 131 (Priest Jesus) Caligiuri 5+5 loop
Cluster 131 (defendant Priest Jesus): Caligiuri signed 5 Rule-20.01 orders on June 5, 2023 and 5 Found Competent orders on September 15, 2023. Cluster total: 220 docket events / 0 incompetency findings.
Holds (exact)
Claim · May 2, 2023 highest-signature day in corpus
May 2, 2023 carries 80 human judicial signatures across the 4,251-document corpus — the highest single day. Browne 40, Mercurio 40. No Borer signature on this day. 60.6% of the entire month's 132 human judicial signatures concentrated on this Tuesday. 100% on 27-CR matters.
Downgraded from 83 (Browne 41 / Mercurio 40 / Borer 1)
Earlier reports cited 83 sigs (Browne 41, Mercurio 40, Borer 1). Re-derivation yields 80 sigs (Browne 40, Mercurio 40, Borer 0). The "60.6% of May 2023" and "100% 27-CR" claims hold exact.
Claim · Mercurio 2023-05-02 13:30 single-slot composition
The Mercurio 2023-05-02 13:30 slot scheduled 117 cases and produced 250 docket events. Composition: 55 Hearing Held Remote, 43 Found Incompetent, 42 Finding of Incompetency and Order, 40 Bail-to-stand, plus a tail of smaller event categories.
Holds (exact, full slot composition)
Earlier reports cited a "28× Found Incompetent + 28× FOI + 28× Hearing Held Remote + 24× Motion + 24× Bail" breakdown; that breakdown is the Howard cluster (cluster 1) subset of the slot, not the full 117-case slot. Page now uses the full-slot composition.
Claim · 43 50+-case slots / 4 owners / 97.7% at 13:30
43 calendar slots in 17 years of MCRO carry ≥50 scheduled cases at one (judge, date, time) tuple. Four officers own all 43: Mercurio 16, Klein 16, Borer 10, Janzen 1. 42 of 43 (97.67%) are stamped at exactly 13:30. The lone exception is Janzen 2025-09-08 at 10:00 in courtroom GC-C1453, with 74 cases.
Holds (exact)
SELECT judicial_officer_norm, hearing_date, hearing_time, COUNT(DISTINCT case_uid) AS n FROM source_hearings GROUP BY 1, 2, 3 HAVING COUNT(DISTINCT case_uid) >= 50;
07 — Regime Precedes Statute

Pre-July-2023 reorganization, with the divergence-rate caveat

Claim · May 2023 = 4.36× January-April baseline
January-April 2023 monthly average: ~30.25 signatures per month. May 2023: 132 signatures. Ratio: 4.36×.
Holds (exact)
Claim · July 1, 2023 core-provisions effective date = null event in top-20 monthly activity
Per-officer monthly time series for the top-20 officers in the corpus shows no change-point firing for any officer at the July 1, 2023 boundary. July 2023 corpus total: 393 events — lower than January, May, or June 2023.
Holds
Cut from main page · Docket-vs-signer divergence (28.2% / 20.0% / 13.6%)
Earlier reports cited a year-over-year docket-vs-signer divergence rate jumping from 0-2% (2017-2022) to 28.2% (2023), 20.0% (2024), 13.6% (2025). Re-derivation against the live database under the natural specification (signer not in docket-officer array) yields 11.3% (2023) / 10.3% (2024) / 0.0% (2025) — same direction (a 2023 jump) but magnitudes off by ~2.5×. The 28.2% figure must come from a different operationalization than the natural one. The page no longer cites the percentages; the qualitative direction (a 2023 jump preceding the statute) holds.
Dropped from main page (numerical magnitudes; direction retained)
08 — Hudleston, Quam, Borer Roster Removal

Per-officer behavioral verifications

Claim · Hudleston 16 events / 4 clusters / 8 on Guertin
Sarah Hudleston appears in 4 of 77 clusters across the forensic set, totaling 16 events. Per-cluster split: Guertin (cluster 1570) 8 events, Crutchfield (153) 6, Lussier (218) 1, Green (307) 1. 8 of 16 (50%) on Guertin alone.
Holds (exact)
Claim · Apr 29, 2025 Guertin reversal: 4 orders, 3 with no recoverable cryptographic signer
On April 29, 2025, Hudleston signed 4 events on Guertin's case (4th R20 order, Forensic Navigator appointment, hearing-event entry, future-judicial-assignment notice). Three of the four events have PDFs attached; all 3 PDFs carry only the MCRO Watermark + a May 3, 2025 administrative-scan timestamp — no recoverable cryptographic signer, no signer_name_raw, no sig_crypto_status, no sig_crypto_valid_flag. The fourth event (the in-person Faretta hearing) has no PDF attached.
Holds (with one nuance: 3 of 4 events have PDFs)
Cut from main page · 2.51× clique-inbound overrepresentation
Earlier reports asserted that Hudleston is 2.51× overrepresented as the inbound destination of Notice-of-Reassignment events from clique members (48.8% from clique vs. 19.5% baseline). The figure is not reproducible against the live database under any of three natural derivations of "inbound NOCR." Closest derivation gave 67.7% Hudleston-inbound vs. 54.5% baseline = 1.24×. The page does not use 2.51×.
Dropped from main page
Cut from main page · Koch → Hudleston → Klein chain at n=8
Earlier reports asserted a Koch → Hudleston → Klein 3-step reassignment chain at n=8. Strict NOCR chain-of-custody yields 0 cases. Looser variant (Koch as prior JO-bearing event, Klein in any later event): 1 case (out of 3 Hudleston-from-Koch). The n=8 generalization is unverified; only the single Guertin instance is forensically demonstrable. The page does not use n=8.
Dropped from main page
Claim · Janzen 2022/23/24/25 = 1,600 / 103 / 115 / 340
Lisa Janzen's annual event count: 2022 = 1,600, 2023 = 103 (-93.6% YoY), 2024 = 115, 2025 = 340. The 2022→2023 inflection is fully inside the seed window. 2024 and 2025 carry the frozen-window flag.
Holds (exact)
Claim · Koch 2022/23/24/25 = 66 / 130 / 408 / 647
William Koch's annual event count: 2022 = 66, 2023 = 130, 2024 = 408, 2025 = 647. 2024 and 2025 carry the frozen-window flag (only Guertin's case has been refreshed since April 2024).
Holds (exact, with caveat)
Claim · Carruthers↔Koch Pearson r = -0.5245
Pearson correlation of monthly event counts between Carruthers and Koch over 72 shared monthly observations: r = -0.5245. Rank-1 most-negative correlation among all top-10 officer pairs.
Holds (with reframe note)
The earlier "annual delta" wording in the source raw_output is a mis-paraphrase: the actual measure is monthly LEVELS, not annual deltas. Page uses "monthly observations."
Cut from main page · Peer absorption multipliers
Earlier reports cited post-statute multipliers vs. pre-statute baseline: Klein 1.98× / Mercurio 1.80× / Browne 1.64× / Koch 1.74× / Lamas 1.89× / Moreno 3.02× / Janzen 0.11×. Seed-window-only re-derivation produces: Klein 2.35× / Mercurio 1.94× / Browne 2.02× / Koch 1.50× / Lamas 2.16× / Moreno 4.16× / Janzen 0.09×. Direction holds (Janzen collapses, peers absorb), magnitudes differ. The page uses the qualitative claim ("the ring balances itself; when one's caseload goes up, another's falls") and the rank-1-most-negative Pearson r.
Cut from main page (qualitative direction retained)
Claim · Quam 3 docket events on Guertin, all clerical
Across his appearances on case 27-CR-23-1886, Judge Jay Quam has 3 docket events — all Notice of Hearing entries (2023-01-26, 2024-10-01, 2024-10-15). Zero Order/Finding/Ruling events.
Holds (exact)
Claim · Quam principal in 3 of 5 highest-volume forensic clusters (not 5 of 5)
Quam's top-5 forensic clusters by event count: Wesley 15, Higgins 11, Johnson 11, Youssuf 9, Taylor 7. Within-cluster ranking: Quam is #1 in Youssuf, #2 in Johnson, #2 in Taylor (3/5). #4 in Wesley. Outside top-5 in Higgins (where Patrick Meade leads with 23). The page uses "principal in 3 of 5" rather than the stronger "always #1 or #2" framing.
Downgraded from "always #1 or #2"
Claim · Quam total corpus appearances = 1,203
Re-derivation yields 746 docket events + 457 previous-hearing attributions = 1,203 combined. Earlier reports cited 1,660; that figure is not reproducible against the live database. Page uses 1,203 where total Quam volume is mentioned (currently no main-page numerical reference; methodology references only).
Downgraded from 1,660
Claim · Borer roster removal (November 28, 2025 capture)
The Hennepin Fourth Judicial District public roster captured November 28, 2025 (90 PDF biographies in /provenance/data/06-fourth-district-judicial-roster/) does not contain a Borer biography. A year prior, he was on the roster.
Holds (file-system fact, not SQL)
09 — Cuts From the Main Page

Findings that survived verification but failed the "five-second test"

The following claims are accurate against the live database. They were cut from the main page because each one requires a schema explanation a non-lawyer cannot read in five seconds. They are preserved here for the schema-literate reader.

Claim · 0.37% event-level JO attribution on "Hearing Held" events
Of 20,307 docket events with event_name matching the four "Hearing Held *" variants (In-Person 11,238 / Remote 4,667 / Using Remote Technology 2,836 / Hybrid 1,566), only 76 events (0.37%) carry a non-empty judicial_officer_norm at the event level. 99.86% of held events recover JO via cross-table join to parents_hearings_previous on (case_uid, event_json__date = hearing_json__date) — the docket-event side is structurally JO-blind.
Holds (exact)
Claim · 6.78% three-source agreement at the (jo, date, time)-tuple level
Across 31,004 case-cells across 15,976 (jo, date, time) tuples derived from source_hearings × parents_hearings_previous × parents_case_events (Hearing-Held filtered, joined to PHP for JO recovery), 2,102 cells (6.78%) show all three sources agreeing. Largest disagreement bucket: PHP+PCE-no-SH at 60.24% (coverage gap from finite SH search seeds). Next: PHP-only at 29.25% (calendar-listed but no docket trace).
Holds (exact)
Claim · "TUA #1 docket event" for the Referee Advisor archetype
All 9 officers in the "Referee Competency Advisor" archetype (Mercurio, Skibbie, Borer, Olson, Larson, Lien, Poston, Bernhardson, Gearin) have "Taken Under Advisement" as their #1 most-frequent docket event. 9 of 9 verified via per-officer event-name distribution against parents_case_events.
Holds (the page uses the plain-English "deferred-decision instrument" phrasing instead of the schema label "TUA")
Caveat · k=3 silhouette score 0.475 for archetype clustering
The three-archetype clustering (Referee Advisor / Rule-20 Signer / Med-Records Gatekeeper) was derived offline via k-means on per-officer event-name distribution vectors. Reported silhouette score: 0.475. Not directly SQL-verifiable (the clustering itself is offline ML). Page treats this as document-attested; the per-officer event distributions that produced the cluster assignments are SQL-verifiable and reproduce.
Document-attested
Claim · Rule-20 Signer cohort signs 59% of corpus R20.01 orders
The 9-officer Rule-20 Signer cohort (Janzen, Klein, Browne, Koch, Lamas, Meyer, Mulrooney, Carey, Quaintance) collectively signs 2,074 of 3,498 corpus-wide Order-Evaluation for Competency to Proceed (Rule 20.01) events = 59.3%. Earlier reports cited "~70%"; the page now uses 59% / "nearly six in ten."
Downgraded from 70%
Caveat · Klein PMH share is 17-18%, not 53%
Earlier reports cited a 53-percentage-point spread between Klein's Probate Mental Health caseload share (53%) and the rest of the Rule-20 Signer cohort's PMH share (0%). Against source_hearings.case_location, Klein's PMH share is 17.5-18%; Janzen and Lamas show 0%; Koch 3.5%; Meyer 0.9%. Direction holds (Klein concentrates in PMH; peers don't), but the dramatic 53-pp number does not reproduce. The page uses "Klein concentrates in PMH; the others don't — yet they share an event-mix archetype" rather than the 53-pp spread.
Downgraded from 53-pp spread
Caveat · Mercurio "31-second clean Guertin signature" framing dropped
Earlier reports asserted that Mercurio's February 14, 2023 signature on Guertin's docket is the only Guertin per-officer signature passing both edits_after_sig_flag and sig_crypto_valid_flag, signed in 31 seconds. No Mercurio sig exists for February 14, 2023 in children_signatures_full_report. The 31-second figure is a parents-side docket-event observation, not a children-side signature observation. Klein, separately, passes both flags on three Guertin orders (2023-06-14 Continuance, 2023-11-15 R20.01, 2024-01-17 FOI). The "single clean Mercurio Guertin signature" framing as written does not survive the children-side spine. The page does not claim it.
Dropped from main page
10 — Source Reports

The forensic PDFs underneath the SQL

The verification work above re-derived every load-bearing claim against the live database. The earlier prose forensic reports that established the underlying analytical scaffolding remain available:

Baseline Report · PDF
MCRO — Judicial Officer Representation Analysis
22 pages · per-officer footprint analysis · 87-name target list
Narrative Synthesis · PDF
MCRO — Judicial Officer Representation Narrative Synthesis
21 pages · clique null tests · archetype derivation
Raw Investigation · MARKDOWN
Judicial Officer Representation — Raw Forensic Output
37-finding deep-dive · SQL-reproducible · raw query log
Companion Page
/77/ — The 77-Cluster Forensic Decomposition
412 cases · 33.6 years · the cluster expansion underlying this page's "77-cluster set"

Anyone reproducing these numbers should run the queries above against ibfmjtwahkwqzcmeyqii directly. Any number that cannot be reached by the documented SQL is a bug; report it.

MNCOURTFRAUD.COM  ·  /judicial/#actors methodology archive  ·  verified 2026-05-09