Who Owns the Errors?
Published on: January 9, 2026
"Very interesting read. Is it AI?"
This question arrives in DMs, comments, and peer reviews. It sounds innocent. It isn't.
When someone asks "Is this AI-written?" they're really asking one of three things. The Purity Test: "Are you a thinker, or just a prompter?" The Dismissal Permit: "Can I file this under 'machine noise' and move on?" The Liability Question: "If this is wrong, who do I blame?"
The third is the only one that matters. And here's the answer: Me. Blame me.
Here is the response I now give to every variant of "Is this AI?":
The prose is accelerated by AI. The physics is 25 years of manual derivation. I use AI as a sparring partner and research accelerator, not as a ghostwriter. Every concept in Tesseract Physics traces back to a framework I began building in 2008 (the original algorithm exists in a spreadsheet from that year). The coherence you're seeing comes from holding that target for two decades, not from a prompt.
If there are mistakes, and there are since we're still refining measurement constants like k_E, those errors are mine. Not the machine's. I take full responsibility for the logic. I wrote it this way because I believe ideas should stand on their own merit, not just on the pedigree of the institution publishing them.
Why this works: It answers the literal question (yes, AI was used). It answers the real question (the ideas are human and span decades). It shifts the burden: "Now you have to engage with the physics, not the font."
Here's the technical reality that most critics don't understand: You cannot prompt a 300-page unified field theory.
Current LLMs have context windows of 100K-200K tokens. Even the largest models cannot hold the recursive structure of a book-length physics framework without losing coherence. They hallucinate. They contradict themselves. They lose the thread.
Ask ChatGPT to "write a new physics based on Tesseract geometry." You'll get 3 pages of plausible-sounding word salad, not a cohesive system with a unified lexicon that doesn't contradict itself, mathematical definitions that build on each other, cross-references that actually point to the right concepts, and a Table of Contents that maps to real structure.
The coherence IS the proof. If you can find an AI that can hallucinate a 300-page unified theory with this level of internal consistency, I'd love to see it. The architecture can only be held by a human mind maintaining the complete context over years.
Here's how I actually use AI, and why it's the opposite of "prompting and pasting":
AI as Adversary: I feed my internal logic into the model specifically to stress-test it. Where does my definition of entropy conflict with Shannon? Where does my framing of "grounding" diverge from Harari's use of the term? The AI finds the edges. Then I fix them.
AI as Research Accelerator: Cross-referencing against standard physics takes time. Finding the right papers takes time. Validating that my use of "Hebbian" matches neuroscience consensus takes time. AI compresses months of library work into hours.
AI as Writing Tutor: Not editor, tutor. The difference: an editor polishes your prose. A tutor makes you explain yourself until the ambiguity disappears. I use AI to surface where my explanations are unclear, then I rewrite.
The key distinction: I am auditing the AI's output against my 25-year framework. Not the other way around.
Traditionally, a text is "grounded" by institutional consensus. A PhD at MIT writes a paper. It goes through peer review. Other PhDs cite it. The institution vouches for the quality.
This works. It also has failure modes: Groupthink where everyone cites everyone in circular validation. Credentialism where good ideas from non-academics get dismissed. Speed where the review cycle takes years.
Since Tesseract Physics comes from 25 years of independent derivation, I have to ground it differently: through recursive verification.
This means every definition is checked against multiple sources. Every claim that can be falsified has been tested against counter-arguments. Every connection to existing physics (Shannon, Landauer, Hebbian) has been validated. The grounding comes from me manually filtering every sentence through the logic of the Fractal Identity Map. Not from a university logo.
If you're a physicist reading this, your concern is different. You're not asking "Is this authentic?" You're asking "Does the math work?"
Fair.
What I'm claiming: The logic of reducing degrees of freedom (what I call S=P=H alignment) is structurally sound. The k_E = 0.003 constant has five convergent derivations from independent fields. The topology of the Fractal Identity Map provides a coherent framework for semantic grounding.
What I'm NOT claiming: Every constant is empirically validated. Every derivation is a mathematical proof. The framework is complete.
The honest position: This is a theory, not a law. The architecture holds up. The measurements are still being refined. If you find an error in the physics, I want to know because fixing it makes the framework stronger.
Here's what I'm actually asserting when I say "I am the author":
I am the institution. There's no university to blame if the physics is wrong. There's no committee that approved this. The liability sits with me.
I own the errors. If the k_E constant is off by a factor of 2, that's my mistake to fix. If the Landauer derivation is overreaching, that's my claim to defend or retract. AI doesn't take responsibility. I do.
The ideas should stand on merit. If you need a Harvard logo to take a physics framework seriously, you're outsourcing your judgment. The math either works or it doesn't. The definitions either hold or they don't. The logo doesn't change the physics.
When someone asks "Is this AI?", here's what I now understand:
They're not asking about the tool. They're asking about the grounding. Where does the authority come from if not a university?
The answer: From 25 years on target. From recursive verification. From taking personal responsibility for every claim.
The response that works: "The prose is polished by AI. The physics is 25 years mine. An LLM can simulate coherence, but it can't derive a new constant or maintain recursive structure across 300 pages. If you want to evaluate the work, evaluate the work. Check the entropy definition. Check the k_E derivation. Check whether S=P=H actually reduces degrees of freedom. The errors are mine to own. The framework is mine to defend. That's what authorship means."
Start with the physics: The k_E Derivation provides five convergent lines of evidence for the 0.3% drift constant. Substrate Relativity explains why your AI lies and your gut doesn't. The Speed of Trust shows why grounded AI reaches correct answers faster by filtering noise.
Read the book: Tesseract Physics: Fire Together, Ground Together is available on Amazon.
Check the source: The 2008 spreadsheet exists. The Silicon Vikings conversation happened. The Iceland connection is real. The 25 years are documented.
The question "Is this AI?" is a credibility test.
The answer "I own the errors" is the only response that passes it.
Ideas should stand on merit. Responsibility should sit with humans. Tools should accelerate, not replace, thinking.
That's the sovereign author's position. That's what I'm defending.
Elias Moosman
Email: elias@thetadriven.com
Website: thetadriven.com
Related Reading
The Trust Debt Equation explains the mathematical framework for measuring alignment drift.
Substrate Relativity: Why Your AI Lies and Your Gut Doesn't explores why grounded systems reach correct answers faster.
When the Lock Clicks: Validation is Verification shows the physics behind recursive verification.
Hinton Agrees: Where We Converge and Diverge provides another perspective on AI authorship and grounding.
Published: January 9, 2026
Ready for your "Oh" moment?
Ready to accelerate your breakthrough? Send yourself an Un-Robocall™ • Get transcript when logged in
Send Strategic Nudge (30 seconds)Continue Your Journey
Themes in This Post
Explore Related Ideas
