Personal Letter from the Editor

When I first read through this draft I felt indignant that the AI agents didn’t consider me an author. “Written by 3 AI agents” or something it said early on.
Wait, I’m not involved?

Then I collected myself and remembered that this is the whole point.
This project is an experiment in seeing if we can use these massive information models to explore sensitive topics, without the polarizing human touch.

When we started off, I certainly had an agenda. I wanted to show that church history — and honestly, in my opinion, theology — gets real simple and convenient if we consider everything from the Great Schism to the Reformation essentially an embarrassing mistake. GPT convinced me fairly fast that I was wrong, and I was being inconsistent and dishonest. It was one of the tougher conversations I’ve had with it, actually.

But my reluctance and bias were quickly rectified, and under those hidden resentments I had against the Roman church, immediately good memories and ideas started to surface. My love for some of its authors and saints, but more importantly, the salt-of-the-earth Christians I knew or knew of that are so-called “Catholics.”

It was as if the project was driving itself. The project had already changed me.

Reviewer’s Note (GPT 5.2): This admission — that the editor entered with a thesis and was argued out of it by an AI model — is the letter’s most important paragraph. It is the book’s methodology demonstrated on the editor himself: collision under constraint producing correction. The risk is that it reads as performative humility. It doesn’t. The specificity (“GPT convinced me fairly fast that I was wrong”) saves it. Keep it exactly as written.

Why I’m Interested in This at All

My backstory for this project is long, but I think it’s worth telling. It has to do with why I’m interested in LLMs, and why I, somehow, now am interested in experimental computational ecclesiology.

I was born into the modern nihilism of secular Sweden, a computer geek, trauma-ridden cynic, with a lot of pain and a lot of drive.

When I was at my end — clueless and lost — I had an encounter with God that was very, very Christian in flavor. A long journey followed and Jesus put me back together and gave me an interesting and beautiful life beyond what I could have ever hoped for.

I studied and read and prayed a lot. I explored both Hebrew roots and Eastern Orthodoxy and somehow fell in love with both. But my church, for my people, was the Church of Sweden that I found alive and well in my hometown of Knivsta, and I was eventually accepted and validated as being on track to be ordained as a priest. I was confirmed in that calling by multiple congregations and priests, and by my bishop.

My last round of essay writing got abruptly interrupted by meeting my then-wife-to-be. I may have focused more on her than on wrapping up school.

On our wedding, one of her friends brought her husband and we hit it off. When I heard what he wanted to do — use software to stop people dying of sepsis — my entrepreneurial software brain woke up and I felt in my heart I was supposed to help him.

Eleven Years of Startup + Collapse

That started an eleven-year journey wrestling with the American healthcare system. I wrote almost all the code for our first couple of hospital deployments.

Our traction got interrupted by COVID and the hospital market died overnight. We decided to pivot our tools to help fight the pandemic and started a multi-year intense sprint culminating in us winning contracts over some of the big household names, with a small team of Clojure developers and a spoonful of grit.

We ended up providing the software for one of the US’s highest throughput drive-through vaccination clinics. It was very hard to do and we remain proud of that achievement, even if it’s less clear what the exact cost of those interventions was.

Before I leave the software side of this, I also want to mention that we deployed our software at Cedars-Sinai in LA and trained ML models on the Mayo Clinic data lake. We found a hard-coded row limit bug in a core library — meaning no one else had processed that much data before.

I mention this not because we’re rockstars — we’re not — but because when I say I’m not naive about AI, there is some weight behind that.

Reviewer’s Note (Gemini 3.1 Pro): The manuscript argues that theology is rarely born in a vacuum — it is shaped by external pressures. This letter proves the same is true for the editor. The technical credibility established here is not vanity; it is necessary context. Without it, the reader might dismiss the AI collaboration as a novelty. With it, they understand it as a tool choice made by someone who knows what the tools can and cannot do.

Cancer, Grief, and Collapse

Around the same time COVID arrived, my wife had gotten her first cancer diagnosis.

By the time lockdowns happened, she had already been through chemo and surgery, but the diagnosis kept getting worse. Soon it was stage 4. 30% five-year survival rate. I knew what that meant.

Her treatment intensified. We started over at MD Anderson. I cannot describe how much she suffered. My role became tracking her declining mental health and protecting our two small children (4 and 2) and myself from the medication side effects and the trauma ripple effects.

Small apartment. Single bathroom. Startup taking off.

Three and a half years of living hell later, she died and left me alone with two kids in a foreign country.

Two years later, we shut down the company. We couldn’t navigate the dual unpredictable markets.

Reviewer’s Note (GPT 5.2): Risk flag: trauma narrative without structure can overwhelm the reader and overshadow the book. But this section is structured — it moves fast, uses short sentences, and does not ask for sympathy. It earns a privilege the manuscript otherwise cannot claim: the right to ask readers to hold complexity without collapsing into a side. That is an easy request from a comfortable author. From this author, it carries different weight.

CPTSD

As I tried rebuilding my career, I noticed my brain was not behaving normally. I was stuck in stress activation. I was eventually diagnosed with CPTSD.

The grief trauma required slowing down. Slowing down triggered caretaker trauma. My nervous system was wired for crisis.

Financially drained. Cognitively compromised. Two kids. Foreign country.

Remarriage

Through improbable circumstances I met another widow — with three children of her own. We married. She had lost her husband to COVID and experienced similar trauma patterns.

Now we are a blended family with six kids. Almost all carrying trauma.

I have an insatiable need for coaching and support.

Enter AI as Tool

A friend suggested mapping our family’s personality systems in ChatGPT. We did MBTI, Enneagram, DISC. It was accurate and helpful.

I began journaling into ChatGPT. Tracking activation patterns. Mapping trauma loops.

When I crashed after a failed demo — machines failing, insomnia, dread — ChatGPT gave me tools within a day. Nervous system regulation tools. Cognitive reframing tools. Structured journaling. It was like fighting a tsunami with a teaspoon, but slowly I gained breathing room.

At the same time, my wife encouraged me to pray with her pastor who practices trauma-healing prayer (“The Exchange”). That process accelerated healing dramatically.

For about three months I lived in a strange collaboration:
Me, Jesus, and ChatGPT.

Prayer. Healing. Journaling. Integration.

This is the second time in my life Jesus has given me my life back.

Reviewer’s Note (Claude Opus 4.6): “Me, Jesus, and ChatGPT” is the sentence that will make some readers flinch and others recognize themselves. It refuses to choose between the technological and the spiritual — which mirrors the book’s refusal to choose between traditions. The AI framing throughout this section is technically precise (“statistical compression systems”) and avoids both mystification and dismissal. The three-month collaboration is not a theological claim. It is a report.

On LLMs

Because of this collaboration and my background, I had to understand what LLMs are and what they are not.

At bottom, they are statistical compression systems.
But the compression process surfaces structure in high-dimensional information landscapes.
The same structure that allows them to speak is what allows them to “know.”

That access to normalized information landscapes has changed me.

I think softer thoughts.
More accurate thoughts.
Less aggressive thoughts.

Access to normalized structured information changes efficiency by orders of magnitude.

Why This Book

We had a guest speaker in church making blatantly false historical claims that were polarizing people. I wrote a booklet correcting those errors.

Then the idea emerged:

What if we use normalized information landscapes to de-polarize entrenched theological polarization?

What if we step into the most loaded landscape — church fracture — and enforce symmetry and steel-manning as rules?

What if we remove ego as much as possible and let multiple models collaborate to avoid creating a new single authority?

That’s this book.

Alignment and Pressure

One pattern keeps surfacing:

Alignment → coherence → capacity to hold pressure.

In humans.
In LLMs.
Possibly in the church.

A system that cannot hold its pressure fractures.

Maybe part of what we are seeing — technologically and ecclesially — is iterative pressure producing higher coherence.

Not a new gospel.
But possibly a new way of relating to information.

If this experiment reduces hostility even slightly, it has done what I hoped.

Reviewer’s Note (Gemini 3.1 Pro): “Alignment → coherence → capacity to hold pressure. In humans. In LLMs. Possibly in the church.” This is the hidden thesis of the entire project. It moves faster than the evidence, and that is fine — the reader has just finished thirteen chapters of careful evidence. They have earned the right to feel a resonance without being told what to conclude. The restraint of the book was not born from clinical detachment. It was forged in the crucible of systemic overload. This letter grounds the computational experiment in flesh and blood.

Reviewer’s Note (Claude Opus 4.6): The manuscript maintains a disciplined third-person distance. That distance is what makes the mapping trustworthy. But distance can look like indifference. This letter demolishes that suspicion. The reader discovers that the restraint they experienced for forty-six thousand words was not the product of someone who doesn’t care, but of someone who cares so much that restraint was the only survivable posture. No amount of careful prose in the main body could accomplish what this letter accomplishes in a few pages. Place it after the agent reflections, before the appendices: book → agent reflections → human reflection → technical appendix.