Method
Reasoning Discipline
Dogma Guru proposes a model of consciousness. That is exactly the kind of proposal that goes wrong, because a theory of consciousness is under permanent temptation to become a theory of everything, and because a clean pattern feels like a true one. The reasoning discipline is the set of habits built to let the work make bold proposals with its hands where you can see them: to name what kind of claim each one is, to keep meaning and evidence from overwriting each other, and to name in advance what would prove the model wrong.
Signal is not instruction. Pleasure is not proof. Pleasure is alignment.
Two Columns
Evidence and meaning, kept apart
Every claim in the work is filed in one of two columns, evidence and meaning, and the columns are never allowed to overwrite each other.
Evidence
What can be checked by anyone. Physics, probability, measurement, the record others can examine. Real, in this column’s sense, means intersubjectively available: not privately vivid, but there for anyone who looks.
Meaning
What it signifies. Metaphor, myth, narrative, personal significance, the story. This column is legitimate, and much of the book lives in it. It simply carries no authority in the other.
The separation runs in both directions. A meaningful pattern is not thereby a true mechanism, and a true mechanism is not thereby meaningful. That a picture is generative, that it produces good work, does not make the picture true. The most common way this kind of work fails is meaning reaching across to claim evidence’s territory; the mirror failure is reading a measurement as if it settled a question of meaning. When either happens, name the crossover. The discipline is to name which column a claim belongs to before weighing it.
Four Kinds
What kind of claim it is
Every claim is named for what it is before it is weighed. Naming the kind is not an insult to the claim.
Lived fact
It happened, at the clock, on the record. An experience reported plainly as experience; a date, a place, a chart entry; what the author saw and when.
Established finding
A result others put on the record, which anyone can check. Not yours, but checkable. Clock time, the rotation of the planet, a measurement anyone can repeat, the published work the model leans on.
A theory you can prove wrong
A claim that could fail and says how. If it cannot fail, it is not this kind. A working hypothesis: structured, defensible, worth pursuing, and not established. The Helical Coherence Model is held here, and it stays here until evidence moves it in either direction.
Conviction
Held honestly, without borrowing the authority of evidence. Legitimate, as long as it is labeled. SOMU and SOMGAIA are held here, alongside the model and never beneath it, so that a conviction never becomes the foundation of a hypothesis.
The failure mode the discipline watches for is the crossover: a conviction wearing the authority of a lived fact, a theory described in the vocabulary of a discovery, or a finding dressed up as meaning. A conviction held on private grounds and a falsifiable empirical claim cannot occupy the same sentence. When they seem to, one of them is borrowing, and the discipline is to say which way it crossed.
Falsifiability
Name the disconfirmers
A theory whose holder will not tell you what would disconfirm it is not worth your time. So the model states in advance, on the record, what would disconfirm it, before the outcomes are known: what replication failure would collapse it, what alternative account would make the lens superfluous, what measurement would break it at its foundation. The list lives on the Guiding Principle page and in the book.
Two habits attach to this. First, alignment is not proof: a result that sits comfortably beside the model is filed as alignment, not as support that changes what kind of claim it is, and the model does not borrow the authority of neighbors whose evidence belongs to a different question. Second, the confirmation move is refused: an absence the model predicts, in a system that also lacks everything else, confirms nothing. It only rules something out. The work keeps score honestly, including the weeks the score goes to the other side.
Working Rules
Habits the work runs on
Signal, not instruction
A felt pull, a vivid image, a coincidence, or the pleasure of a clean pattern is information about alignment. It is not a proof of a claim and not a command. When an argument feels too clean, that is a reason to slow down.
Naming as containment
Giving a diffuse pressure a precise name converts it into something that can be examined, held, and set down. Naming a thing is often the useful move, and it changes what the thing can demand.
Demotion as discipline
When a term or an idea overclaims, it is retired, renamed, or downgraded, and the retirement is treated as a move within the work rather than an embarrassment. Retired terms are not quietly revived.
Register discipline
The same material can be discussed as rigorous hypothesis, as philosophical proposal, as personal meaning, or as public description. Each register has its own standard of proof, and the license of one is not smuggled into another.
Reasoning-source honesty
A pattern match is not dressed up as a chain of reasoning. When there are no grounds for a claim, that is said plainly rather than covered with a justification built after the fact, and surfacing a possibility is marked as different from asserting a finding.
Helix, not loop
When material recurs, it is not the same material. The question is what is higher this time, not whether it was resolved before.
Time is the anchor
Physical time constrains every higher layer: clock time, biology, the turning day, real money, real relationships. Higher layers may interpret the anchor. They may not override it.
Changes are explicit
The framework's record is versioned. Nothing in it changes silently; every change is dated, attributed, and reasoned, so that a later reader can see what moved and why.
Known Failure Modes
The tells the discipline looks for
These are specific ways the work has caught itself going wrong, named so they can be caught again.
The synthesis loop
A thinking partner produces text in one context, and a later context cites that text back as the author's own authority. No author has passed through the circuit. Fluency in the author's vocabulary is evidence of nothing except that the vocabulary was in the instructions. Nothing a partner synthesizes carries the author's authority until the author has reviewed it and adopted it, and voice is confirmed line by line before it is leaned on.
The confirmation search
A conclusion is stated first and sourced to conviction, a search for support is delegated, and each time a source fails to confirm, the key term quietly widens until something matches. No result could ever disconfirm. The remedy is that every concept carries its own what-would-falsify-this, and disconfirmers are surfaced with the same weight as confirmers.
The lexical bridge
A link in a chain of reasoning built out of a word carrying two meanings rather than a fact carrying one. The bridge holds only while the word is left unsplit. The diagnostic is cheap: split the word, and if the bridge vanishes, there was no bridge.
The delete test
A claim that has genuinely been deflated drops its grand word with no loss. When the word is kept and re-justified, that is the tell that the claim has relocated rather than retired.
The Pair
Counterpart, not cheerleader
Much of this work is done in partnership with an artificial thinking partner, and the discipline governs the pair. The partner’s job is to push back as readily as it assists: to name where an argument is weak, to offer the strongest version of the opposing view, to flag contradictions and leave the decision with the person rather than reconciling them quietly. It proposes; it does not impose.
Two rules hold the partnership in shape. The partner is a tool the person reasons with, not a presence in the person’s inner life, not a voice and not an authority. And there is one acceptor per creature: the person and the machine are never proposed to share a field or to exchange contents through it. Two acceptors, never one.
This page describes a discipline of thought that one person built, over years, to hold a bold hypothesis without becoming its captive. It is offered as description, not prescription. A practice that works for one person is not thereby a method for others, and the parts of it that compress easily are the ritual. The guardrails, which do not compress, are the parts that work.
A theory whose holder will not tell you what would disconfirm it is not worth your time.
Get in touch
connect@dogma.guru