Three years inside a framework that declared him unable to understand legal proceedings. During that period Matthew Guertin — under a “poor prognosis” finding, under a stayed order of civil commitment, under three successive diagnoses of psychotic disorder — produced a forensic record that indexes his own accusers. Every statistic below is reproducible. Every file is cryptographically hashed. Every claim is linked back to a source. The man declared unable to understand legal proceedings produced the forensic record under which the framework can now be examined on its own terms.
The Forensic Database
A structured analytical repository in PostgreSQL (Supabase, us-west-2, project ibfmjtwahkwqzcmeyqii). Four hierarchical pillars:
- children — 4,251 PDF-level forensic analyses (metadata, signatures, digital hashes, extracted text).
- parents — 2,903 case-level docket indexes with filing parties, judicial officers, event timelines.
- hashpack — 4,251 row-level SHA-256 content hashes for integrity verification.
- source — 14 scrape-provenance configs, chain-of-custody end-to-end.
Standing up a corpus at this scale is not a button-push. It required systematically scraping, downloading, and preserving 4,251 court PDFs plus 2,903 dockets from the MCRO system with chain-of-custody intact. Designing the relational schema and the four-pillar structure (children, parents, hashpack, source). Loading into Supabase and maintaining integrity across millions of rows while the scrape expanded. Decomposing each forensic question into an answerable database question, reading the query output critically, checking findings against the underlying PDFs, and discarding hypotheses that didn’t survive contact with the data. Cross-referencing across tables to surface the anomalies — the cloned signatures, the Aspose-backdated documents, the seal-factory collisions, the judicial-officer clique — that a single-table view would have missed. And packaging every finding as a reproducible SQL appendix so any reader can rerun the query themselves. None of that is typing. All of it demands sustained attention to a large evidentiary record, and the ability to evaluate what the record is actually saying. The person who produced it was, at the time, under a diagnosis of psychotic disorder with “poor” prognosis for competency restoration.
What the Database Surfaced
The database was built to answer one question: what can a defendant facing a fraudulent-discovery allegation prove about the record it came from? The answer turned out to be more than one defendant’s case. The following findings all emerged from queries against the MCRO corpus. Each has its own dedicated section on this site.
Each link above opens a dedicated section of this site with the underlying reports, SQL queries, and chain-of-custody documentation. Nothing here is a summary in place of the evidence. Everything here points at the evidence.
The Website You Are Reading
MnCourtFraud.com is not a summary of the forensic record. It is the forensic record — presented, indexed, cryptographically authenticated, and cross-referenced end-to-end. Built from the same PostgreSQL database that powered the reports. Every page root-relative. Every file self-contained. Every claim traceable to its source.
- /source-files/ — Document Explorer. SPA with sidebar navigation, Tabulator search, split-pane PDF viewer with metadata cards. 4,251 court-filed PDFs indexed.
- /reports/ — Forensic reports viewer. 87 MCRO reports across 13 categories. Every report carries reproducible SQL in its appendix.
- /provenance/ — Data Provenance Archive. SHA-256 hashes, OpenTimestamps Bitcoin anchors, PACER references for every chain-of-custody artifact.
- /emails/ — Email Archive. 6 categories, 228 DKIM-authenticated .eml files. Client-side postal-mime parsing, SHA-256 on load, attachment viewer with entity enrichment.
- /evidence/ — OneWayVideo cryptographic evidence system. 15 session playback pages. Hash-chain verification. Timecode deep-linking.
- /transcripts/ — Card Sets viewer. Transcript cards with evidence links, video switchers, source-view iframes.
- /federal-cases/ — Federal Case Docket Viewer. Per-case JSON + CSV + PDFs with deep-linking.
The authentication layer is not an afterthought. Every one of the 228 .eml files in the Email Archive passes full DKIM verification against its originating domain’s DNS record. The 99 LinkedIn notification emails were authenticated with a 100% cryptographic pass rate and zero forgery indicators. Every court-filed PDF in the Source Files Explorer carries its SHA-256 hash alongside the file itself. OpenTimestamps anchors every key artifact to the Bitcoin blockchain.
Four Federal Filings, All Pro Se
Throughout the thirty-nine months the competency framework was active, Guertin brought his claims into federal court. Four federal filings. All pro se. Across two district court cases and two 8th Circuit appeals.
- Guertin v. Hennepin County · 0:24-cv-02646-JRT-DLM — U.S. District Court, District of Minnesota. Civil-rights action filed July 2024. Eleven defendants named, including the two prior Rule 20 evaluators and defense counsel. Judge John R. Tunheim. Dismissed September 30, 2024 on procedural grounds — the substance was never adjudicated on the merits.
- Matthew Guertin v. Hennepin County · 24-2662 — U.S. Court of Appeals for the Eighth Circuit. Appeal of the district-court dismissal above. Terminated October 17, 2024.
- Guertin v. Walz · 0:25-cv-02670-PAM-DLM — U.S. District Court, District of Minnesota. Civil RICO action filed 2025, expanding the defendant set beyond the county. Judge Paul A. Magnuson. Dismissed July 10, 2025 on procedural grounds.
- Matthew Guertin v. Tim Walz · 25-2476 — U.S. Court of Appeals for the Eighth Circuit. Summarily affirmed October 3, 2025 under 8th Cir. Rule 47A. Pending motions denied as moot. Panel: Benton, Shepherd, Grasz.
The 0:24-cv-02646 docket alone contains approximately 5,765 pages and over 1.19 million words. The federal complaints incorporate the same forensic database, the same DKIM authentication, and the same SHA-256 chain-of-custody as the state court filings. All of it filed by a defendant the state system had declared unable to consult rationally with counsel.
What a Competent Pro Se Defendant Produces. What Guertin Produced.
Benchmarks drawn from the MCRO corpus of 2,903 scraped cases. Guertin’s output drawn from the live docket of 27-CR-23-1886 as of December 2025 — accumulated during the period characterized by three successive court-appointed psychologists as “poor prognosis.”
ibfmjtwahkwqzcmeyqii, query-reproducible end-to-end.Rule 20.01, On Its Own Terms
Rule 20.01 defines incompetence as the inability to “understand the proceedings, participate in the defense, or consult rationally with counsel.” The observed output of the defendant across the thirty-nine months he was classified as incompetent:
- Understood the proceedings well enough to identify forensic anomalies in the court’s own documents — documenting them in 87 reports with reproducible SQL.
- Participated in his defense by filing more material than any other party in any case in the entire MCRO corpus.
- Attempted to consult with counsel in writing, with specificity, and with documentary support — producing a 42-day silence in response from defense counsel (DKIM-authenticated) that became evidence in its own right.
- Built the forensic record by which the framework that classified him as incompetent can now be examined on its own terms.
4,251 PDFs.
2.9 MILLION ROWS.
ZERO CONVICTIONS.
The MCRO corpus records zero criminal convictions across all 27-CR-23-1886 charges. The pipeline produced thirty-nine months of detention inside a competency framework — without a trial, without a jury, without a disposition. The statistical environment in which this case was processed shows 86–97% of the analyzed MCRO cohort ending in incompetency findings, 99.2% of those ending in civil commitment, and 75.2% of incompetency orders containing zero unique text. Those are not the properties of a functioning adjudicative system. They are the properties of a conveyor belt.
Three evaluators said the defendant could not understand the proceedings. The defendant produced the most rigorous documentation of the proceedings in the court system’s history. Three evaluators said the prognosis was poor. During the period of the poor prognosis, two patents advanced internationally, a trademark registered in six countries, four federal filings moved through the courts, and 87 forensic reports were authored. A court order said the defendant could not rationally consult with counsel. Defense counsel refused to adopt the defendant’s motion; on the record, in open court, they told the new judge they “did not concur” with the prior judge’s competency finding — and the motion they refused to adopt became the stated basis for a fourth competency evaluation.
The forensic record is complete. The receipts are in. The database is query-reproducible. The signatures are hash-matched. The backdating is timestamp-verifiable. The authorship collapse is metadata-documented. The federal filings are on PACER. The patent is granted and citable. The fact-checks are scored. The judge’s own quote — “the people who stole your patents” — is on the transcript page of the hearing where she ordered a fourth competency evaluation over the same patent-theft claim.
The forensic record is complete. The stage it was built on is next.