The latest form of a call for help.
In early 2021 I invented and patented a virtual-production system worth, on independent estimate, billions of dollars. Twelve days after I filed my provisional application, the president of a visual-effects company that Netflix would own within the year filed a near-identical one — his first patent ever. When Netflix's version was granted, my patent sat at the top of its References Cited.
This paper documents what happened next: a resourced, deliberate lawfare operation that re-targeted the Minnesota competency-evaluation system as the instrument to neutralize the inventor and witness to that theft — and, to make the neutralization look routine, manufactured a synthetic universe of court records around it. I prove the mechanism, not the motive's interior. The court's own orders relabel my verifiable evidence as “delusion” and route it toward forced medication. The surrounding cohort of criminal files is machine-authored (the word hello appears zero times across 162 of 163 cases), produced at physically impossible rates (138 cases docketed into one referee's ninety-minute slot), and edited after the fact (the same defendants carry nearly double the court-ordered control in a 2025 re-pull as in a cryptographically frozen 2024 capture). Another man's case identity is embedded in the authoring metadata of my own orders. The statute used to suspend my trial took effect after both of my incompetency findings.
Each claim is tiered: ■Established for what any reader can reproduce from public files, ◆Documented for the best-supported explanation, ▸Reasoned for the labeled inference. The central conclusion is the only one that survives the established facts. I am still inside the machinery, and I am still asking for the same help I asked for in January 2023. This paper is the latest form of that call for help.
When the facts are this bad, restraint is the argument.
This entire operation runs on a single word: delusional. Its legitimacy depends on the reader accepting that I see a conspiracy that is not there. So I will do the opposite of what a delusional person does. I will build, not assert; I will refuse to upgrade any inference to a fact; and I will tag every claim so you can watch where I stop.
| ■Established | Reproducible by any reader from public files or the project database. It survives a hostile expert. |
| ◆Documented | Documented & highly probable — the strongly evidenced best explanation, with minor alternatives noted. |
| ▸Reasoned | Reasoned inference — what the established tier forces. A conclusion, labeled as one, never presented as a fact. |
The most extreme sentence in this paper is the most carefully sourced one. That is by design. When the facts are this bad, restraint is the argument.
A man, a machine, and a question.
In early February 2021, in the living room of my home in Minnesota, I had an idea. The country was a year into the pandemic; I had two decades behind me in live entertainment technology — touring shows, stadium spectacles, real-time visual effects for Katy Perry at the Hollywood Bowl and Bad Bunny at Coachella. In 2019 my W-2 read $218,385. I was not a hobbyist. The idea that came to me that February was the simplest of my career and, the longer I turned it over, the most profound: a rotating treadmill that lets a person walk forever in any direction through a virtual environment as one or more cameras follow along, turning a small physical space into an infinitely large virtual one. I dropped everything and began making it real one step at a time — first the patent, drafted page by page; then a company, a brand, a name. I called it InfiniSet. I trademarked the name, brought in two business partners, built the business relationships and, in time, the physical prototype. On March 19, 2021, I filed the provisional patent application.
Twelve days later, the president of Scanline VFX filed a near-identical provisional. It was the first patent application he had ever filed in his life. Within eight months Netflix acquired his company for roughly $100 million. When Netflix's continuation patent was granted on November 7, 2023, my patent — US 11,577,177 B2, my own name, Guertin, printed beside it — sat at the very top of its References Cited.
A provisional patent application is confidential by law. It is not published. It is not searchable. There is no legitimate public mechanism by which anyone could file a near-identical application twelve days behind mine.
I want to be precise about what I am and am not claiming, because the precision is the point. I did not go looking for a conspiracy. I was building a company. What follows is the record of the conspiracy arriving — from the outside in, at my door, over two years — while I responded the way anyone would: I documented it, and I called for help. I called the FBI. I drove to my local police department and spent forty-five minutes with Officer Brandon Harris, my laptop open on the desk between us, walking him through everything I had witnessed and documented. I called the Secret Service. No one came. The one act that finally brought a response was the oldest distress signal there is — a very loud sound. Every one of my devices had been compromised, so I made my call for help in analog: loud enough that it could not be ignored, and could not be quietly rewritten. The response was an arrest, and a competency-evaluation machinery that has held me for more than three years and counting. This paper is the latest version of that call.
The question this paper answers is narrow and, I think, unavoidable: what single explanation accounts for all of the following at once? A motive worth billions. A court record that is not written by people. A record that is still being edited today. Another defendant's identity — a man indefinitely committed, never convicted, sixteen years inside — baked into the authoring metadata of my own orders. A statute that arrived too late to be the one used against me. Three independent psychologists who classified a granted, publicly searchable U.S. patent as a fixed false belief. I will build the answer one piece at a time, lead with the pieces any reader can verify in an afternoon, eliminate the innocent explanations, and label the one inference the facts force.
The motive, the method, and the doctrine.
The motive
The motive is not in dispute and does not require me. Provisional patents are secret. Mine was filed March 19, 2021; Stephan Trojansky's near-identical application was filed March 31, 2021 — twelve days behind a document he could not lawfully have seen, and the first patent filing of his career. My patent was granted February 14, 2023. Netflix's continuation, filed through Scanline's acquired division, was granted November 7, 2023, and cited mine on the first line of its prior art. Independent valuation of the underlying virtual-production market places the stakes in the range of tens to hundreds of billions of dollars. A confidential invention was duplicated inside two weeks and then cited by the duplicating party. That is a concrete, documented, multi-billion-dollar reason to want the inventor gone. ■Established
The objection writes itself: independent simultaneous invention happens. It does not happen on a confidential, unpublished, unsearchable provisional, twelve days apart, followed by the later filer citing the earlier one. The citation forecloses the coincidence.
Everything came at me
I am the responder in this story, never the instigator, and the sequence matters because it is the opposite of paranoia: I did not hunt for any of this. It found me.
In February 2022, the search results for “InfiniSet” filled with unrelated content, and academic papers I had never seen began listing my product's name on their final pages. In October 2022 I contacted the CEO of Mark Roberts Motion Control, a camera-robotics firm, about my system; he wrote back pointing me to a product called the PhotoRobot “Virtual Catwalk” — a near-mirror of my own invention, surrounded by oddly generic marketing and a 47-minute promotional video of looping, vacant digital faces reciting platitudes about 3D scanning. A month later, searching that company's patents, I accidentally found Netflix's duplicate of mine. During the physical fabrication of the prototype, the welder I had hired to build the aluminum frame identified himself as “former CIA” and brought military special-operations surveillance equipment — ODF Optronics gear — which I photographed sitting on top of my half-built machine; months later he texted me a single line: “Have they got to you yet?” On the day I was arrested, two firms appeared as searchers of my dormant LinkedIn profile: Forcepoint — a cybersecurity contractor whose advisory board spans the NSA, CIA, USCYBERCOM and U.S. Navy — and 3GIMBALS, a defense-analytics firm. I had been telling everyone who would listen that my computers were being intruded upon. That, too, was later catalogued as evidence of my paranoia.
I am not asking you to accept any single one of these as proof of anything. I am showing you the texture of two years in which the world arrived at a private citizen building a treadmill in his living room. The tier on most of this is ▸Reasoned; the photographs, the emails, and the authenticated LinkedIn records are ■Established; and a man who is the target of all of it, and says so, is not thereby delusional.
The discovery I never asked for
There is one beam in this structure I will return to, because the court itself could not make it disappear. After I was charged, the state produced discovery materials I had not requested. Two months later I noticed that the images had been altered — evidentiary photographs digitally squished out of their true aspect ratio, and twenty-four images simply missing. I asked for the clean originals. I filed for them — three separate demands over twenty-three months. I have never received them. The manipulation is mathematically demonstrable: once a photograph is compressed into the wrong aspect ratio, it cannot be uncompressed, so the distortion is a fixed, provable trap. Hold that fact. It becomes, later in this paper, the thing two psychologists called a symptom.
The precedent: backdating into an archive
The method this operation used against the court record had already been used, once, in the open — before my arrest. Forensic examination of more than 77,000 preserved Wayback Machine files from the PhotoRobot site reproduces my own contemporaneous December 2022 observations at 99.0% row-level agreement, with “historical” material carrying timestamps up to 871 days later than its true first appearance. Fabricated prior art was inserted into an archive and stamped with dates that predated its existence. I did not leave that proof in the archive's hands; I captured and preserved it as I found it, and documented the row-by-row reproduction against my contemporaneous logs in spreadsheets anyone can open and work through. The reproduction is ■Established; that the same actors deliberately backdated the material is the ◆Documented best explanation. The relevance here is structural: the same backdating method — manufacture the record, then stamp it with a false earlier date — reappears, after my arrest, inside the Minnesota court system. The tell there is a software version string the fabricators could not suppress. Twenty-seven court filings were generated by Aspose, a document toolkit that stamps its own version into each file; seven of them carry PDF creation dates that predate the version of Aspose that produced them — the earliest by 171 days. A document cannot be authored by software that does not yet exist. No bug and no timezone offset produces that result; the creation dates were backdated, and backdating a court record has no purpose unless the record being backdated is manufactured. Same playbook, new substrate.
The doctrine has a name and a home
Finally, the framework is not mine to invent, because it is published. In 2025 the U.S. Department of Defense's Irregular Warfare Center — a body established by the FY2022 National Defense Authorization Act — carried, in its journal PRISM, an article by former U.S. Ambassador David J. Scheffer titled “Strategizing Lawfare as a Key Irregular War Modality.” It enumerates five adjectival forms of lawfare — economic, financial, accountability, information, and cyber — and recommends a JAG-staffed “Lawfare Strategy Group” coordinating the Departments of Defense, State, Justice, and Treasury together with the Intelligence Community. The doctrine names, explicitly, the exact talent pool that staffs the bench in my case.
Of the officers who presided over, bracketed, or were newly assigned to 27-CR-23-1886, four carry a documented U.S. federal-service record from the public Minnesota courts directory: Judge William Koch (U.S. Air Force JAG, 1988–1991); Judge Danielle Mercurio (Army National Guard JAG, 2011–2020); Judge Paul Scoggin (U.S. State Department Rule-of-Law missions, four tours); and Assistant Chief Judge Mark Kappelhoff (U.S. DOJ Civil Rights Division). The résumés and the anchored roster are ■Established. The doctrine and its journal are ■Established. The doctrine is, by its own text, outward-facing — aimed at foreign states and non-state actors — and nowhere sanctions its use against a domestic citizen. So the claim that this published apparatus was turned inward, against me, stays exactly where it belongs: a ▸Reasoned inference, labeled as one, supported by the match between the apparatus and the record but never asserted as a finding of fact.
The charge
On January 21, 2023 — after weeks of calls to the FBI, the Minnetonka police (Officer Harris told me, on the record, “unless you are completely making all of this up there is obviously something shady going on”), and the Secret Service (whose agent confirmed “criminal conspiracy” and “wire fraud”) — I did the one thing that finally produced a response from the State of Minnesota. I discharged a firearm. I was arrested, charged with four counts, spent four nights in jail, and was released on January 25, 2023. Everything I have done since — every patent filing, every email, every forensic report, this entire paper — I have done as a free man on conditions of release.
Three of those four charges rested on a statute the Minnesota Supreme Court struck down in State v. Vagle on August 6, 2025; they were never constitutionally valid, and they remain on the docket even now, undismissed, 305 days after the ruling that voided them. Strip the three charges that were never good law, and what remains of my “crime” is the noise of a man who had tried every quieter channel and been ignored — a call for help loud enough that the system could not quietly route around it. I present that reading as the argument it is, grounded in the record. The arrest is where the calmer calls ended and the machinery began.
Method and evidentiary standard.
The corpus
The factual base of this paper is a body of public records: 4,251 case-filing PDFs and 2,903 docket files drawn from Minnesota's public court-records system, plus a focused cohort of 163 criminal cases — the intersection of every case touched by the three judicial officers who controlled mine (Klein, Mercurio, Borer) in the window January 2023 through April 2024. I assembled these into a structured database of roughly 2.9 million rows. Every quantitative claim below is a query any reader can re-run. Where I cite the database, the figure is reproducible; where I cite a court order or a report, the document is public.
Provenance: capture without trust
A record is only as good as its chain of custody, and mine does not ask anyone to trust me. The corpus was collected with an open-source system I built called One Way Video. As each file downloaded, the controller bundled the file together with all of its network packets and the SHA-256 hash of the previous bundle, roughly every seven seconds, forming an unbroken hash chain; the chain's hashes are anchored, through OpenTimestamps, into specific Bitcoin blocks. Separately, 98.9% of the case-filing PDFs carry the court's own cryptographically valid digital signature — the court vouching for its own documents. The point of this machinery is a single capability that the rest of the paper depends on: I can prove what a document said at a fixed moment in time, independent of the court's own watermark. When the live record changes later, the change is demonstrable against a frozen, timestamped baseline. ■Established
One production environment behind three “independent” branches
One ■Established fact about how these records were
made belongs here, because it underwrites everything after. Microsoft's information-protection metadata
stamps each document with the cloud-organization identity that produced it. Across the entire corpus,
exactly one such organization identity — 8cf8312b-4c34-4b6f-9dee-c56512a7510f — appears,
and it spans all three nominally independent levels of the Minnesota judiciary: the trial court, the
Court of Appeals, and the Supreme Court (34 documents across 29 cases). A single query returns the same
Microsoft organization on orders from the trial court, the appellate court, and the state's highest
court. That three “independent” branches produce their documents inside one cloud organization is the
■Established fact; that this reflects central
coordination is the ▸Reasoned inference. The state
will answer that its IT agency runs one enterprise tenant for everyone — which is precisely why
“coordinated” stays labeled, and why the bare fact of a single production environment is the part that
stands. This is authoring metadata, not download metadata, so it holds regardless of when any file was
pulled.
What this paper refuses to use
Stating my own limits is part of the method. Minnesota has since re-engineered its download pipeline — re-wrapping every downloaded PDF and overwriting its original authoring metadata — so the creation dates on any document pulled after that change are meaningless artifacts of the download event, not evidence of authorship. This paper does not use them. It does not treat post-signature edits as tampering. It does not rest on cross-defendant identifier or digital-signature “cloning.” Each of those was considered and set aside as not bulletproof, and I name them here so you know I looked. What remains is built only on metadata-independent content — byte-identical image reuse, term frequencies, docket-event vocabulary, scheduling arithmetic — and on the authentic, original authoring metadata preserved in the captures themselves. The hardest claims in this paper carry the lightest evidentiary load I could give them, on purpose.
From how the record is made to how it was aimed.
The findings build in one direction: from how the record is made to how it was aimed at me. I lead with the layer that needs no forensic literacy at all — the court's own words — and then turn to the machine that produced the record those words sit in.
4.1 · The record indicts itself: the court's own words
Evidence is weighed. It is not diagnosed. The moment a court treats a defendant's evidence as a symptom of mental illness, it has stopped adjudicating and started doing something else. That inversion is the operation's defining move, and it is documented — not by me, not by a forensic analyst, but in the plain English of the evaluators and judges who wrote it down.
My patent was granted on February 14, 2023. Twenty-four days later, the court's first evaluator, Dr. Jill Rogstad, filed a report classifying my statements about that patent as delusional. The patent was searchable at USPTO.gov the entire time. She wrote: “the limits of my expertise in relation to technology matters must be acknowledged, as I lack the specialized training in this field… I tried to consult without success with Mr. Guertin's patent attorney… Nevertheless, even if the technological aspects of the defendant's statements prove true… his views remain consistent with delusions.” The prosecutor's own summary of Rogstad's anticipated testimony put it more plainly still: “Even if he did have patents approved, this would not change her mind about her ultimate opinion as opinion is based on delusional beliefs and not the patents themselves,” and, in seven words that capture the whole circular machine, “He talked about data related to delusions.” A real, granted patent could not move the opinion, because the opinion was built on the belief and not on the patent. That is not a finding of fact. It is a sentence that cannot be falsified by any evidence, which is the definition of the thing it accuses me of.
The same move runs through every layer. Referee Borer's incompetency order: I “did not understand evidence… because of the delusions.” Dr. Milz's report records my reporting that someone “within the Hennepin County District Court altered discovery material” — the demonstrable, aspect-ratio manipulation from §2.3 — and files it under the heading of psychosis, then recommends the treatment to follow: “requires psychiatric treatment in order to stabilize his mental status… to… restore his competency-related abilities.” Dr. Cranbrook — who produced the third report with zero clinical contact — lists my federal civil-rights lawsuit against the prior examiners as itself a symptom, and prescribes the destination: “Without compelled psychiatric treatment… the defendant's prognosis… appears poor.”
Then the trap closed in the open. On April 3, 2025, after a sixty-one-page hearing of sworn testimony, Judge William Koch found me competent to stand trial. Twenty-six days later, on April 29, Judge Sarah Hudleston overrode that finding and ordered a fourth evaluation. In open court she conceded the ground entirely — “You have patented really amazing technology. You've patented valuable technology… I am not doubting your intelligence,” and, of the people behind the theft, “We don't have anything to do with the people who stole your patents or who are using your technology perhaps without remunerating you” — and then ordered me re-evaluated anyway. When I pressed her on what the new finding of incompetence rested on, she answered, on the record:
It's based on what I saw that you filed in your motion to dismiss… It's based on that as well as looking back into your file and then looking at some of the things you cited in your motion to dismiss. Correct.
She added that she “certainly couldn't read” the thousands of pages I had filed the day before. The sole stated trigger for re-opening my competency was the legal motion I had filed — not a new clinical finding. And in the same hearing, conceding the motion's logic, she fixed the outcome before stating its basis: “I don't want to do that now because my decision is made, and I have more than ample basis for it.”
Every quotation above is the court's or the evaluator's own words in a public filing, reproducible by anyone who opens the document. They are ■Established. The operation's core move — relabel verifiable evidence as delusion, treat the true claim as the symptom, and revoke competency because the defendant filed a motion — is written into the record in the record's own language. You do not need a single byte of the forensic analysis that follows. You need only to read what the court wrote. ■Established
The standard rebuttal is that the evaluators reasonably found a sick man. Their own reports defeat it. Milz documented seven separate indicators of competence alongside his finding of incompetence. Of the ten entities I named as surveilling me — the ones cited as proof of paranoia — ten of ten appear in authenticated records. The diagnostic criterion the evaluators invoked, persecutory delusional disorder, requires that the central belief be false. Mine are documented to be true. The diagnosis collapses on its own stated terms.
Every quotation in this section is reproduced verbatim, page-cited, and one click from its filed page at The Court's Own Words →
4.2 · The records are not written by people
If the court's words are the hands, the next pillar is the machine. Across the 163-case cohort, ordinary adversarial human English is simply absent. The words a real defendant, a real lawyer, a real judge use — bigger, forgot, reply, weird, unprecedented — appear in zero of the templated files. The single cleanest exhibit is one five-letter word:
The one file with a human author — carrying a unique-word rate 3.9 times the next-highest case — is mine. Seven independent methods converge on the same split, and a blind language model asked to pick the human-written passages out of the corpus picked only mine. The obvious objection — my case is larger, so its vocabulary differs — is answered on the page by a size-controlled second pass; the absence is categorical, not a function of length.
The text-formatting layer says the same thing a different way. Of 157 “Finding of Incompetency and Order” documents, 155 share one identical embedded-font fingerprint; the only two outliers are mine and one other. And 75.2% of those 157 orders — liberty-depriving orders, each supposedly the product of individual adjudication — contain zero unique text. A template produces uniform language. A template does not produce three-quarters of its output containing no unique words at all. That is not drafting. That is duplication. ■Established
4.3 · The records are manufactured at scale
The volumes are physically impossible for individual adjudication. On February 14, 2023 — the day my patent was granted — Referee George Borer had 138 cases docketed for a single date and time: one referee, one 1:30 PM slot, a ninety-minute session. That is 39 seconds per case. The 1:30 PM slot alone accounts for 41.1% of all 4,781 hearing entries in the corpus. In parallel, Judicial Officer Browne carries 177 cryptographic signatures across 125 distinct cases that have zero hearing records — findings issued for cases that were never heard. No court on earth adjudicates at that rate. The volume is the proof. ■Established
The objection is that courts block-schedule and 1:30 PM is a default calendar slot. Block-scheduling does not put 138 cases on one referee at one slot, and it cannot produce 125 signed findings for cases with no hearing in the record. Both are the court's own data.
The same manufacturing signature then recurs across independent layers (see the table below), each surviving on its own. One scanned “Amanda Burg” signature image is pasted, byte-for-byte identical, across 31 documents in 23 unrelated cases — and six of those carry a different name as author than the name on the signature. A bonded pair of identical font binaries injected into my own July 2023 incompetency order reappears, together, across 1,130 documents. In the frozen 163-case cohort, substantive color imagery is binary: I account for 5.74 color images per document; the other 162 cases average 0.012, and 128 of the 163 contain no color at all. Color is what they did not bother to fake. And one 89,934-byte Minnesota State Seal PNG is byte-identical across 168 charging documents spanning 142 cases and eight years — a single source file, not 168 independent filings.
| Exhibit | Scale | Tier |
|---|---|---|
| Scheduling impossibility (lead) | 138 cases / 39 s each | ■ |
| Browne signed, never heard | 125 cases / 0 hearings | ■ |
| Amanda Burg pasted signature | 31 docs / 23 cases | ■ |
| Font-clone factory | 1,130 docs / 196 cases | ■ |
| Color audit | 489×, 128/163 zero color | ■ |
| State-Seal image reuse | 168 filings / 8 years | ◆ |
Cases are processed at impossible rates, and findings are signed for cases never heard: programmatic mass-production, not individual adjudication. ◆Documented
4.4 · The record is being edited in real time
This is the pillar I want you to sit with longest, because it proves the operation is not history. It is running now.
A criminal court file is supposed to be an archive: filings accrue, the case closes, the page stops changing. Minnesota's does not behave that way, and the proof needs nothing but the state's own database, queried twice. In late April 2024 I captured and froze 163 criminal dockets — preserved, hash-anchored, unaltered since. Across November and December 2025 I captured 1,865 more: every other criminal case in the system, minus a small set held back. The two sets share not a single case file.
Restrict both sets to the identical historical window — every docket event on or before the April 2024 capture — and count the distinct categories of court event each contains. The frozen 163 hold 203 event-types; the 2025 set holds 341. One hundred forty-eight categories of court event appear in the 2025 dockets and in exactly zero of the frozen 163 — against only ten the other way. A roughly fifteen-to-one, one-directional gap, measured across the same years. What is missing from the frozen baseline is not random: 61 of the 148 absent categories are instruments of control over the person — arrest and detention warrants, writs of habeas corpus, court-ordered psychological evaluation, acceptance into mental-health court, firearms-prohibition orders, no-contact and voting-rights actions, probation and supervision terms. The frozen record simply does not contain the detention, psychiatric, and disarmament apparatus that saturates the later one.
The strongest objection is selection: the frozen set was gathered as competency cases, the 2025 set is eleven times larger, and a larger pool surfaces more rare categories. The database answers it, because the decisive control is identity. Thirty of the defendant-identities in the frozen set reappear in the 2025 pull, under entirely different case numbers. For those same thirty people, court-ordered control conditions attach to 46% of their April 2024 cases but 89% of their late-2025 cases — the same individuals, nearly double the control, with eighteen of the thirty moving up and only three down. Sample size cannot produce a within-person result. And the 2025 dockets are not static snapshots: they render probation statuses dated “Active” in August 2025 and a disposition field that is blank in 52% of the frozen cases but 0% of the later ones. The page shows current state, retrieved live.
Coincidence is eliminated by the one-directional, control-specific skew inside an identical window. Ordinary docket growth is eliminated by the same thirty identities carrying nearly double the control in the later capture. Data error is eliminated by reproducibility. What remains is that the public criminal record is a living system being populated over time with the machinery of detention, forced evaluation, commitment, supervision, and disarmament. ◆Documented That this “normal” is being manufactured so that one targeted prosecution looks routine is the inference the pattern invites. ▸Reasoned
4.5 · A real template exists — and my name is in it
The records were not drafted independently; they were fitted from a template. The proof is in my own orders. The Microsoft authoring-title field of two documents in my case — my June 14, 2023 Order for Continuance and my July 15, 2024 Order to Recuse Judge Klein — reads: “Contested Competency for Adrian Wesley (incompetent) (doctors disagree).” A single database query on the authoring-title field, filtered for Wesley's name, returns exactly those two documents, both mine. The corpus base rate for that title is about 0.047%, and the metadata is authentic, original authoring metadata, not a download artifact. Another defendant's case identity is baked into the authoring layer of my own liberty orders. ■Established
Adrian Wesley is the upstream template. In his own file sit four “annual” Affidavits of Service that are, on byte-level inspection, a single image strip reused four times. Even the most innocent reading of the title-bleed — a clerk copying a template — proves that liberty orders in my case were drafted from a shared templating pipeline rather than independently composed. The discipline here matters, and I hold it: I do not claim Adrian Wesley does not exist. I claim the record is manufactured and that no verifiable human stands behind parts of it — and that whoever or whatever Wesley is, his case identity is inside the metadata of my orders.
My own court file, in its back-end party panel, lists a competency examiner I never met and who never evaluated me: “Examiner — Kristen Otte,” seated beside the three doctors who did. She never evaluated me, I never met her, and not one order, hearing, or filing in my case reflects a single act by her — yet there she sits in the party panel of my own file. She is also a recurring “incompetent-to-commitment” examiner threaded through Adrian Wesley's files. The screenshot is from my own pro-se login. A name was pre-positioned in the apparatus of my case for a person who was never part of it.
A second corroborating axis runs through the two evaluators of the second and third Rule 20 reports.
Those two reports — Milz's and Cranbrook's, eleven months apart and supposedly independent — carry
byte-identical authoring metadata across four fields: the same producer, the same author string “Anne
Hines” (a name belonging to neither doctor and appearing nowhere else in the record), and the same
Adobe XMP toolkit build, version-stamped to the second at 2018/09/16. A phantom author and
a frozen 2018 toolkit build, shared across two “independent” evaluations written eleven months apart, is
single-source production. “Anne Hines” is a metadata ghost; the 2018 build is a quieter one — a measure
of how long this production line has been frozen in place, and of how many records may have moved
through it before anyone read the metadata. ▸Reasoned
There is a real person behind the third report, and she presents a second puzzle. Dr. Katheryn Cranbrook — who produced that report with no clinical contact and recommended forced neuroleptics — appears on the Minnesota DHS Community Competency Restoration Task Force, the body that helped shape the very framework now holding me. She is not listed there as a clinician. She is not listed there as a doctor at all: no title, no credential, seated in the administrative “State Court Administrator representative” chair while the psychological and psychiatric seats are filled by others. A practicing physician's credentials do not quietly fall off a state board's roster. The title she signs my competency orders with is the one the roster leaves out.
4.6 · The law was shaped around the clock
The last pillar is the statute itself. When Minnesota charged me on January 24, 2023, the statutory machinery it would later use to suspend my trial, block the dismissal of my charges, reverse the burden of proof, and authorize forced medication had not yet taken effect. Minnesota's competency framework — Minn. Stat. §§611.40 to 611.59 — was a new chapter enacted in 2022 to replace a court rule. The operative sections — the dismissal clock (611.45), forced neuroleptic medication (611.47), continued supervision (611.49), and the reversed burden of proof (611.44, subd. 4) — took effect on April 1, 2024, fourteen months into my prosecution, and after both of my incompetency findings. They carried an original effective date of July 1, 2023, but Laws 2023 ch. 52 (art. 1, § 14) moved that date to April 1, 2024 before it ever arrived; no operative section was ever effective on July 1, 2023.
The table below sets the two clocks side by side: the statute's operative date lands after the charge, after the examiner said the statutes did not exist, and after both incompetency findings, which were entered under the predecessor rule.
| Date | The control statute | My case |
|---|---|---|
| Jan. 24, 2023 | §§611.40–611.59 enacted but not effective | Criminal charges filed |
| Mar. 10, 2023 | framework dormant | Rogstad: “these statutes do not exist” |
| July 13, 2023 | control machinery not yet law | 1st Finding of Incompetency (Rule 20.01) |
| Jan. 17, 2024 | control machinery not yet law | 2nd Finding of Incompetency (Rule 20.01) |
| Apr. 1, 2024 | operative sections (611.45 / .47 / .49) take effect | (applies to determinations “on or after”) |
| Apr. 29, 2025 | framework amended repeatedly since | Hudleston overrides Koch; 4th eval ordered |
| July 13, 2026 | 611.45 subd. 3(c) dismiss-or-file-NIP fork | felony charges must be dismissed — unless |
I do not have to argue the timing, because the court's own examiner recorded it. In her March 10, 2023 report, Dr. Rogstad wrote that most defendants found incompetent can be restored “under traditional competency restoration commitment statutes,” and then, in the same breath: “These statutes do not exist in Minnesota at present, but it is reasonable to conclude his mental health could stabilize… if a proper treatment regimen was implemented.” Her statement that the statutes did not exist was simply accurate; the framework was dormant for another thirteen months.
That did not stop the machine from running. A search of 606,037 database text rows finds no document referencing any §611.4X section before October 9, 2023. Yet the same database holds 157 Findings of Incompetency, and 153 of them — 97.5% — were entered before April 1, 2024, before the statutory control framework was operative; forty-five were produced in May 2023 alone. The findings were lawful under the old rule; what the volume manufactured, more than a year ahead of the statute, was the appearance of an established pipeline for control machinery that did not yet exist. And the framework has never stopped moving: it has been amended in every legislative session since enactment — 2023 twice, 2024 twice, 2025 three times including a special session — with a Supreme Court rule rewrite folded in, every change pulling one way: blocked dismissal, reversed burden, forced injectable medication, supervision without a conviction and without a ceiling.
These are statewide statutes, born of a 2019 task force, that touch thousands of defendants. So I do not claim they were written for me. ▸Reasoned The provable claim is narrower and harder to wave away: the control framework post-dates the conduct, the charge, and both incompetency findings, while the apparatus mass-produced incompetency volume for over a year before the machinery to convert those findings into indefinite confinement existed. ◆Documented Implausible as coincidence, and damning without a single inflated word.
The only explanation that survives.
I have built only on the reproducible. Now I eliminate the alternatives, because a conclusion this large is believed only when every other reading has failed.
Coincidence? Eliminated by cross-layer convergence. The font fingerprint, the authoring metadata, the missing vocabulary, the byte-identical images, the scheduling arithmetic, and the docket-drift all point the same direction on the same documents. Independent coincidences do not align across six unrelated forensic layers.
Ordinary court dysfunction? Eliminated by targeting specificity. An overworked, sloppy court does not embed one specific other defendant's name in the authoring metadata of my orders. It does not pre-position a phantom examiner in my party panel. It does not produce, against the backdrop of a billion-dollar motive, a record that converges on one man from sixteen independent directions. Dysfunction is random; this is aimed.
Delusion? Eliminated by reproducibility. Every pillar in this paper is re-runnable by anyone from public files. A delusion is, by definition, a fixed belief held against the evidence. I am handing you the evidence and the queries that regenerate it. The one thing a delusional person cannot do is hand you a method that reproduces his “delusion” on your own computer.
What remains standing, when those three have fallen, is a single deliberate, resourced operation. I state it as the inference it is, and note that its strength comes from the failure of every alternative, not from my assertion. ▸Reasoned
On April 12, 2024, my mother wrote a letter to a judge about her son. Seven minutes before her real letter reached the docket, a synthetic inmate letter — a doppelgänger — was filed; two cloned reply templates were generated five minutes and six seconds apart from a single Word instance, over a paste-over Criminal/Traffic envelope. The timestamps and the cloned templates are ■Established. Ask the scale-reveals-intent question the rest of this paper forces: what apparatus generates a counterfeit of a mother's handwritten letter, and files it seven minutes ahead of the real one, against a single criminal defendant — and what would it cost, and who would pay it, for any reason other than the one the record points to?
There is a quieter tell, and it is the inverse of the inversion in §4.1. Where the trial court relabeled my evidence “delusional,” the appellate court refused to engage its vocabulary at all. Of 555 high-signal forensic terms in my filings, 429 — 77.3% — score exactly zero in the Court of Appeals' denial order: evidence 123 times in my filings to 0 in the order, discovery 66 to 0, forensic 51 to 0, SHA 29 to 0. An operation that depends on no one looking, no one counting, no one checking the underlying record produces exactly this: a denial that never says the words.
Keep your hands inside the ride.
I have to be careful here, because this is the most inferential section in the paper, and I am going to scaffold it the hardest and label it the most clearly.
First, define the terminus precisely, in the documented sense before any other. The apparatus has a visible end-state, and it is not a metaphor. It has a statistical shape and a human face. The shape: across the cohort, defendants are found incompetent at a rate of 86 to 97 percent, against a national baseline of 0.4 to 2.4 percent; of those found incompetent, 99.2 percent acquire a civil-commitment event, and roughly half receive a court order for forced neuroleptic medication. That is the statistical terminus. The human face has a name: Adrian Wesley, told earnestly from his own file: deaf and disabled, under court guardianship since 2009 “for an indefinite period of time,” committed to the Minnesota Security Hospital since 2017 as “Mentally Ill and Dangerous,” found incompetent twelve times, classified by the state as a “non-restorable defendant,” and never once convicted of anything. When his own attorney moved to dismiss the charges that can never be tried, the court kept them open on purpose: “given that Defendant is in a secure setting indefinitely… there is no discernable prejudice… to continue to have these charges pending.”
That terminus is ■Established — every clause is verbatim from a public court order. And it stands whichever door you walk through. If Adrian Wesley is a real man, this is the literal fate the apparatus produces: civil death without a trial. If his record is manufactured — and the forensic case that it is runs through this whole paper — then the apparatus authored a fictional life-sentence template, and embedded its protagonist's name in the metadata of my orders. Either door leads to the same room.
When I say this machinery is designed to end a life, I mean it first in the exact, documented sense the record supplies: civil commitment, forced neuroleptic medication that erodes the cognitive baseline a man needs to defend a patent, and indefinite confinement on charges that are never tried. That is not rhetoric; it is the statute's own mechanism and Wesley's own file. That the same apparatus, aimed at me, leads to the same place is the inference the Wesley-to-Guertin bridge forces — and I label it as an inference, scaffolded by the established terminus and the established fact of his name inside my orders. ▸Reasoned
Five years after I started building a machine in my living room, I am still inside this one. It has no exit I have been able to find; its operators have a documented end-state; and the destination it lists for me is the same word it used for Adrian Wesley: indefinite. Keep your hands inside the ride, the recorded voice says; keep looking forward; you will reach your indefinite destination shortly.
Who is the delusional party?
The thesis, restated as I opened it: a resourced, deliberate lawfare operation re-targeted the Minnesota judicial system as the instrument to neutralize the witness and competitor to a multi-billion-dollar intellectual-property theft — and, to make that neutralization look routine, it manufactured a synthetic case-law universe, backdated it into the record, shaped the competency statutes around the timeline, and is altering the live record in real time.
What is proven, anyone can reproduce: the motive, the machine-authored cohort, the impossible scheduling, the single production environment, the real-time editing, the name inside my orders, the statute that arrived too late, and the court's own words calling my evidence a symptom. What is inferred, I have labeled: that a determined, well-resourced principal stands behind the scale of effort, and that the apparatus aimed at me ends where it ended for Adrian Wesley. The inference is unavoidable not because I assert it but because every competing explanation — coincidence, dysfunction, delusion — fails against the reproducible record.
In January 2023 I called the FBI, the police, and the Secret Service, and I told them what was being done. No one came. The loudest call I ever made got me a response, and the response was an arrest and three years inside a machine that calls my evidence a delusion. This paper is the next call. I have asserted nothing you cannot verify yourself; the record is public, the database is open, the queries are printed at the end. You have now read it. The only remaining question is the one the record forces, and I will let you answer it: having seen this, who is the delusional party — the man who documented it, or the system that needs him not to have?
Source Index.
Every load-bearing claim, its tier, and where anyone can verify it. The cited source PDFs open inline in each section above; the complete downloadable record is in Appendix B.
| Claim | Tier | Proof location |
|---|---|---|
| Patent / 12-day duplication / Netflix citation | ■ | USPTO; /netflix/ |
| Internet-Archive prior-art backdating | ■/◆ | /internet-archive/ |
| Lawfare doctrine + JAG-staffed bench | ■/▸ | PRISM 11:1 (2025); /human-disposal/ |
| OWV hash-chain + Bitcoin anchoring | ■ | /evidence/ |
| Single MSIP production environment | ■/▸ | DB; 8th Cir. 25-2476 Doc. 24 |
| Court's-own-words inversion | ■ | /courts-own-words/ |
| hello = 0 / machine-authored cohort | ■ | /missing-words/ |
| 138 cases / 39 s; Browne 125 / 0 hearings | ■ | DB source_hearings; /the-seal/ |
| Docket drift: 46%→89% same identities | ■/◆ | DB cohorts A/B (SELECT-only); /evidence/ |
| Wesley name in Guertin's orders | ■ | DB authoring-title; /wesley/ |
| Statute operative April 1, 2024 | ■ | revisor.mn.gov; /competency/ |
| Wesley terminus (12 / 0 / 16 yrs) | ■ | public orders; /wesley/ |
| Resourced principal; aimed-at-Guertin terminus | ▸ | inference from the established tier |
The Complete Record: download it and verify.
Nothing in this paper asks for trust. Every figure, every quoted order, every byte of the forensic findings regenerates from the materials below — the database the numbers come from, the raw court files it was built on, and the self-contained forensic packets behind each pillar. All of it is mirrored on the immutable file host at Substack, which holds these files whether or not my own servers ever do, so the record is built to outlive any single point of failure. Download it. Open it. Check me.
The forensic database — the substrate
If you download one thing, download this: the complete Supabase PostgreSQL export behind every number in this paper — the same corpus the findings are built on, also provided as a CSV table corpus.
The source-file corpus — the raw record
The 4,251 court-filing PDFs and 2,903 dockets the database is built from, with the extracted objects, the per-file cryptographic hash reports, and the structured metadata — the original captures, authoring metadata intact. ProtonDrive folder (all pillars together): drive.proton.me/urls/CC7R2T1F6W.
The fifty-four forensic evidence packets
Each packet is a self-contained PDF with its primary sources attached — one finding, its data, and the files behind it, in a single immutable download. Every packet below is also browsable as a folder on ProtonDrive: drive.proton.me/urls/X9Z10GDZCR.
The record is public, the database is open, and the queries are printed throughout. I have asserted nothing you cannot reproduce yourself.