
Quadratic Logic: Building the Human-AI Scribe From the Documentary Record
OUTPOST 422 | JEDVIDENCE™ | JED422-RESEARCHER
After years of collecting emails, administrative filings, employment records, agency determinations, correspondence, timelines, screenshots and firsthand notes, I discovered something important: the story isn’t contained in any single document.
It exists in the relationships between them.
That observation became the foundation of Quadratic Logic, the analytical framework behind my developing JEDVIDENCE™ methodology and the Jaded Patriot Brief.
The experiment asks a deceptively simple question:
What happens when a human journalist and an AI scribe repeatedly examine the same documentary record from four different directions?
The four corners are:
Lived Experience + Documentary Evidence + External Corroboration + AI-Assisted Variable Analysis.
None of those corners is permitted to become the truth by itself.
My memory isn’t the record. An employer’s position statement isn’t the record. An investigator’s determination isn’t necessarily the entire record. And ChatGPT certainly isn’t the record.
The documents remain the sources.
The Human-AI partnership instead operates as a scribe—organizing, comparing, indexing and repeatedly testing those sources while the human remains responsible for interpretation, verification and publication.
That distinction is fundamental.
Accuracy Through Convergence
The goal isn’t to create an artificial intelligence that tells readers what happened.
It is to develop a workflow capable of showing readers why a particular proposition can—or cannot—be supported by the available record.
A source enters the system as a PING.
The Human-AI workflow examines dates, speakers, claims, documents, contradictions and corroborating material. JED422-RESEARCHER can then compare those findings against statutes, administrative rules, judicial decisions and other authoritative materials.
The resulting organized evidentiary product becomes the PONG:
JEDVIDENCE™.
Accuracy therefore isn’t based upon trusting AI.
It comes from making the AI show its work against human-controlled source material.
The 18 Recurring Variables
Scanning my employment and administrative matters as a combined dataset has exposed recurring questions. They are not findings that every employer, attorney or government official committed wrongdoing. They are research variables—patterns sufficiently recurrent to justify systematic examination.
The eighteen themes currently emerging from the dataset are:
1. Accommodation communication.
Who knew about a disability-related limitation, when did they know it, and what happened afterward?
2. DVR involvement.
When vocational-rehabilitation professionals were involved, what communications existed between DVR, employers and the worker?
3. Notice gaps.
Were people or organizations who appeared important earlier in the chronology included when later decisions were made?
4. Documentation versus recollection.
Do contemporaneous emails, messages and records support later descriptions of events?
5. Timeline compression.
Do later narratives combine separate events in ways that change their apparent meaning?
6. Retaliation chronology.
What protected or complaint-related activity preceded an adverse action, and what intervening events occurred?
7. Escalating rhetoric.
Did terminology describing an employee become more severe over time, and what evidence accompanied that change?
8. “Threat” characterization.
When conduct or speech was described as threatening, what precisely was said or done, who witnessed it, and what contemporaneous evidence exists?
9. Witness foundation.
Was a factual proposition based upon firsthand observation, hearsay, inference or later reconstruction?
10. Missing witnesses.
Were potentially corroborating or contradicting witnesses interviewed, identified or otherwise represented in the record?
11. Missing physical or digital evidence.
Did potentially relevant video, email, scheduling, personnel or other electronic evidence exist, and was it examined?
12. Progressive discipline.
What coaching, warnings, improvement plans or disciplinary steps existed before termination—and what did the employer’s own policies contemplate?
13. Decision-maker separation.
Who supplied information, who characterized it, and who actually made the employment decision?
14. Military-service language.
When military background entered workplace discussions, exactly what was said, by whom, and in what context?
15. Administrative narrative versus source record.
Does an investigative determination accurately reflect the material evidence submitted to the agency?
16. Unresolved documentary conflicts.
When two records materially disagree, does the administrative analysis acknowledge and resolve the conflict?
17. Procedural completeness.
Were the theories actually raised by the complainant investigated and addressed under the governing procedural framework?
18. Record provenance.
Can every important proposition ultimately be traced backward to its original source?
These variables don’t determine the answer.
They determine where we look.
That’s Where Quadratic Logic Changes Journalism
Traditional narrative journalism often moves forward:
Event → Interview → Story.
JEDVIDENCE can move forward, backward and sideways.
A termination letter can be compared against an earlier email. That email can be compared against DVR documentation. The resulting discrepancy can be checked against another witness’s account. The chronology can then be tested against Wisconsin employment law and administrative procedure.
Quadratic Logic keeps asking four questions:
What did I experience?
What does the documentary evidence establish?
What can an independent source corroborate?
What patterns emerge when AI examines the variables across the complete dataset?
Agreement among those corners increases confidence.
Disagreement is equally valuable.
A disagreement tells the researcher where another investigation is required.
The Human Remains in Command
This may be the most important lesson from the experiment.
AI should not become the witness.
It shouldn’t become the judge.
It shouldn’t quietly transform an allegation into a fact because the allegation appeared repeatedly in the dataset.
That is why Outpost 422 increasingly labels information according to epistemic status: documented fact, allegation, perception, inference, disputed proposition, legal argument or independently verified finding.
The Human-AI scribe becomes useful precisely because it can preserve those distinctions.
And when it gets something wrong, the human corrects it and sends the corrected proposition back through the analytical loop.
That makes JEDVIDENCE less like asking a chatbot a question and more like maintaining a continuously tested research notebook.
From Four Employment Cases to a Research Laboratory
My individual matters provided the source material, but the experiment has become larger than any single dispute.
Marcus Palace Cinema, Frank Productions, Camp Createability and Madison College contain different parties, different circumstances and different procedural histories.
They should not be collapsed into one allegation.
Quadratic Logic instead asks whether recurring structural questions can be examined consistently across independent matters.
Who knew what?
When?
What document proves it?
Who witnessed it?
What changed?
What evidence contradicts it?
Was the contradiction addressed?
What rule governed the decision?
And can another researcher reproduce the analysis?
Those questions are becoming the grammar of the Human-AI scribe.
Journalism From the Trenches of Metadata
That is what we’re building at Outpost 422.
Not an AI oracle.
Not a machine that declares winners and losers.
Not software that turns suspicion into fact.
We’re developing a Human-AI evidentiary scribe designed to preserve chronology, provenance, contradiction and uncertainty while helping a human researcher navigate documentary volumes that would otherwise be extraordinarily difficult to compare manually.
The ambition is substantial:
PING → Quadratic Logic → JED422-RESEARCHER → Human Verification → PONG → JEDVIDENCE™.
Every iteration adds another opportunity to test the methodology.
Every discrepancy becomes another variable.
Every correction strengthens the provenance trail.
And every source remains available for the most important question investigative journalism can ask:
Show me the record.
Outpost 422 — Stoic Journalism From the Trenches of Metadata™









