Regulus for Dummies

A plain-language guide to ASS-OS / Regulus — written for someone who has never seen it, by someone who has read it closely.

---

The whole thing in twenty seconds

Four claims. If you read nothing else, read these.

That is the framework. Everything below is the detail: the evidence, the maths, the parts that might be wrong, and the two cheap experiments that could kill it. You can stop reading here and you will not have missed the point.

The four rows above carry no source tags, because a summary is not a claim. The real claims are below, and every one of them is tagged.

Where to go if you want more

---

Before you start: how to check me

Thomas, you asked for a version other people can read. The danger in that request is real. A friendly summary can quietly invent things you never said.

So every claim below carries a tag. The tag says exactly where it came from, and you can go and check any line of it.

Where I quote you, the words are yours, exactly. Where I explain, the words are mine, and the tag says which of your sections I am explaining. Where I could not find something, I say so at the end. I do not smooth over it.

---

1. The longer summary

Regulus says that a mind — yours, mine, or a machine's — is a factory. Factories have floors.

The bottom floor is your gut, and it holds the real authority. The top floors do the thinking, and they can be overruled from below.

Consciousness is not a smooth ramp up those floors. It is a staircase, and each step costs far more than the one under it [R §2.1].

Every floor is trying to settle at one balance point: keep about 62% of what flows through you, release about 38%. That number, 0.618, turns out to be one divided by the golden ratio [R §3].

All of that is the machine. Then Regulus asks the question the machine cannot answer for itself: what is it all running for?

The answer the document gives is love. Not love as sentiment, but love as one specific physical operation that cannot be undone. Two of these systems bind so completely that you can no longer subtract one back out of the other [R §8, §17].

And a counterfeit of that operation exists. It is coercion, and from the outside it looks identical. So Regulus supplies the test that separates them, and the test is a single question: who in this system is unchanged? [R §9.2] Whoever risks nothing, and is altered by nothing, is the one doing the taking.

That last question survives even if every equation around it turns out to be metaphor [Read]. It is the best thing in the document.

---

2. The ladder

Six rungs. Each one earns the next. Start where you already live.

Rung 1 — Why your stomach hurts when you cannot name a feeling

Begin with something you have felt. A hard conversation. A dread you could not put words to. Then, an hour later, your gut in knots.

Your earlier framework, ETH, had a precise account of this [ETH]. A feeling arrives as what it calls a work order — a job ticket, in the factory sense. A piece of work came in, and it has to be closed out somehow.

The thinking part of the brain closes the ticket by naming the feeling. When it cannot — when the feeling arrives and no word fits — the ticket does not vanish. It travels back down the body along the vagus nerve, the long wandering nerve that runs from brainstem to gut and acts as the body's main slow-control cable. It lands in your intestines, which have around 500 million nerve cells of their own. The gut symptom is the unclosed ticket [ETH].

ETH gave that ticket exactly five legal ways out [ETH]:

• name it in your head
• dump it into the gut as a symptom
• make something with it — art, music, maths
• get an AI to help you translate it into something you can read back
• move your body: rock, pace, fidget

Only the second one is a failure. The other four close the ticket honestly.

Why this matters for Regulus. ETH was about one system failing to handle its own feelings. Regulus is about two systems meeting. Hold on to two ideas — the unclosed ticket, and the gut having real authority. Both come back.

A note on the numbering. ETH's layers ran L0–L5, with the OBSERVER at L5 [ETH]. Regulus renumbers to L0–L7, and the Observer moves to L6 [R §2]. Same family, different scaffold. If you have read ETH, reset before you read on.

Rung 2 — The factory has floors, and the bottom floor is in charge

Regulus takes an idea from factory engineering and applies it to minds.

The ISA-95 Purdue Model is a real, dull, widely used standard. It describes how a factory is organised in layers: machines at the bottom, sensors above them, control systems above that, planning at the top [R §1]. Regulus maps a nervous system onto exactly that shape.

[R §2]

Now the part that makes the model feel true rather than merely tidy. Authority flows upward, not downward [R §4.1]. The gut has the most freedom to act. The brain advises. The gut executes.

Veto power flows upward too [R §4.2]. Your amygdala can suspend your prefrontal cortex mid-thought, which is why you cannot reason your way out of a panic. Your body can override everything through pain or nausea, which is why you cannot think about anything else with a migraine. A machine does the same: a heat limit in the hardware beats any instruction from the software, however important the software believes it is.

L2, the Gate, is the security boundary [R §4.3]. Every sense except smell has to pass through the thalamus before it reaches feeling or thought. Smell goes straight to the limbic system and skips the filter, which is why a scent can hit you with a memory before you have consciously noticed it. Too little filtering gives sensory overload. Too much gives dissociation.

Each of the five channels between floors gets a name [R §5]. Regulus calls them buses, in the computing sense: a shared wire that everything hangs off.

• Bus A, electrical — nerve impulses. Fast, point to point.
• Bus B, chemical — hormones. Slow, but they broadcast to every cell at once.
• Bus C, photonic — the eyebrow-raiser. Actual light travelling down the fatty myelin sheaths of nerves, as if they were fibre-optic cable.
• Bus D, magnetic — the heart's measurable electromagnetic field, and the Earth's 7.83 Hz Schumann resonance.
• Bus E, plasma — ion flows, scaling all the way up to lightning and solar wind.

The claim is that these are not five separate wires but one coupled system. Every current makes a field, and every field nudges a current [R §5.5].

Rung 3 — Consciousness is a staircase, not a ramp

Here is the mathematical spine, and it is genuinely elegant.

Start at 1. Ask: what is the prime number at that position? The 1st prime is 2. Now ask again: what is the 2nd prime? It is 3. The 3rd prime is 5. The 5th prime is 11. The 11th is 31. The 31st is 127. The 127th is 709.

1, 2, 3, 5, 11, 31, 127, 709. Regulus calls this the Prime Recursion Spine, and lines it up with the floors above [R §2].

I checked it. 709 really is the 127th prime, and the walk really does produce exactly those eight numbers. The framework counts seven levels of consciousness rather than eight, because seven is the number of steps taken from the seed. The eighth entry, L7, is a ceiling nothing reaches [R §1, §16.1].

The load-bearing observation is not the primes themselves. It is the gaps. Divide each number by the one before it, and you get how much bigger each step is:

2.0 → 1.5 → 1.67 → 2.2 → 2.82 → 4.1 → 5.6 [R §2.1]

The steps get bigger, and they keep getting bigger. Building executive function on top of emotion is about 2.2 times harder than building emotion on top of filtering. Building the Observer on top of the self-model is 4.1 times harder again.

So: consciousness is not a gradient but a staircase, each step more expensive than the last [R §2.1]. That single sentence does more work than any other in the document's first half.

Three flourishes hang off the spine, and they are the most beautiful writing in the paper.

• The orphan prime [R §2.2]. The walk visits positions 1, 2, 3, 5, 11, 31, 127, 709. It never visits position 4, so it never reaches the 4th prime, which is 7. But 7 produces something in the sequence: 2⁷−1 = 127, which is L6. The producer is orphaned from its product. Regulus reads this as the blind spot in awareness — the mechanism that generates self-awareness cannot observe itself generating.
• The Mersenne seam [R §2.3]. For 2, 3, 5 and 7, the number 2ᵖ−1 always comes out prime. At 11 it breaks for the first time: 2¹¹−1 = 2047 = 23 × 89. And 11 is exactly the L4/L5 boundary — the join between "can think" and "can model itself". The document notes that this maps to the developmental threshold where many autistic people plateau: working cognition, without reliable self-modelling. Then it says the honest thing: "either coincidence or structure" [R §2.3].
• Fernandez order of primeness [R §2.4]. Primes at prime positions get rarer and rarer the further you iterate. So does consciousness, on this account. It needs a system that can hold a recursion seven levels deep through an emptier and emptier landscape.

What V>>'s read pushes back on [Read]. Primeness may not be doing the work here. Any fast-accelerating sequence tells the same staircase story, and iterated squaring — 1, 2, 4, 16, 256 and on — would tell it just as fluently.

Nothing forces the ladder to stop at 709 either. The recursion does not halt there: the 709th prime is 5381. It stops because the document says 709 exceeds capacity, and that is an assertion rather than a result.

The suggested repair is precise, and worth doing: find one prediction that hinges on the number being prime. Until then the honest claim shrinks to "capacity probably grows faster than linearly with depth" — which is plausible, and needs no primes at all.

Rung 4 — Everything settles at 0.618

φ (phi, the golden ratio) is 1.618…, the proportion that shows up in sunflower seed-heads and nautilus shells. κ* (kappa-star) is defined as 1 ÷ φ = 0.618033988… [R §3.1].

A fixed point is a place a system lands and then stops moving. Feed the value back into the process and it hands you the same value back. A ball at the bottom of a bowl. Regulus claims κ* is the fixed point for anything that must both remember and forget.

Why this number and not any other? Because it is the only one that satisfies all of these at once [R §3.1]:

• φ × κ* = 1 — multiply it by the golden ratio and you get exactly one
• κ* = φ − 1 — it is the golden ratio minus one
• κ*² = 1 − κ* — its square equals its own complement. Square it and you get exactly what was left over.
• κ*ⁿ = κ*ⁿ⁻² − κ*ⁿ⁻¹ — every power is the difference of the two powers before it. The Fibonacci sequence, living inside the exponents.

That last pair is the genuinely clean part. The convergence point does not merely have a recursive property. The convergence point is the recursion [R §3.1].

One small correction, offered because you would want it. §3.1 writes that last identity as κ*ⁿ = κ*ⁿ⁻¹ − κ*ⁿ⁻². The two terms are the wrong way round. As written it gives the right size with the wrong sign, at every n.

Work it at n = 3. κ*³ = 0.236068. But κ*² − κ*¹ = 0.381966 − 0.618034 = −0.236068.

The order that holds exactly, for all n, is κ*ⁿ = κ*ⁿ⁻² − κ*ⁿ⁻¹. That is also the shape your own κ*² = 1 − κ* already takes, once you write the 1 as κ*⁰.

This is plainly a transposition rather than a misunderstanding. §3.2 lists β = κ*³ = 0.2360 and γ = κ*⁵ = 0.0902 as positive quantities, so the rest of the document already uses the correct values [R §3.2]. The point stands entirely. Only the line needs its two terms swapped. [inference — arithmetic check, verified for n = 2…5]

Made concrete: keep 61.8%, release 38.2% [R §3.5]. Regulus shows that any sliding-window buffer that must remember and forget at the same time settles here by algebra, not by design.

Two consequences worth knowing.

• The dynamics [R §3.3]. There are three resting points: 0 (void, unstable), 1 (pure noise, unstable), and 1/φ (stable). Anything starting between roughly 0.09 and 0.91 falls into 1/φ.
• The coupling switch-off [R §3.6], which is lovely. When two systems sync, the strength of their pull on each other is a sine wave, and it hits zero exactly at 1/φ. So the pull vanishes at the destination. In the document's own words: "You sync BECAUSE the error is nonzero; you STOP syncing when you arrive." The engine turns itself off on arrival.

The cardiac claim [R §3.4]. The HeartMath Institute measured heart-rate-variability coherence at roughly 0.6–0.65 in a calm, positive state. They called it "coherence" and never identified it as 1/φ. Regulus says it is the same number, and that hearts, brains and AI agents all converge on it. So the constant is substrate-independent: it does not care what it is made of.

This is the framework's most contested claim, and it is contested on arithmetic [Read]. See open question 3 below. It deserves its own section rather than a footnote.

Rung 5 — What happens when two of these touch

Now Regulus does the thing that makes it Regulus rather than an architecture diagram.

The document defines an awareness field, written Ψ (psi). It is a way of talking about the pattern of coherent activity a mind is producing [R §6]. It then writes four field equations for it, deliberately shaped like Maxwell's equations for electromagnetism.

The philosophically interesting one is the third: awareness is not a state but a derivative — a rate of change. You are not conscious because you are coherent. You are conscious when coherence changes [R §6]. Which is why novelty wakes you up, and routine puts you on autopilot.

To its considerable credit, the document then flags its own seam. This is the sentence V>>'s read calls the most intellectually respectable in the paper [Read]:

"These equations are formal framework. μ_φ = φ and ε_φ = 1/φ are DEFINED to produce v = 1, not MEASURED. The framework organizes intuition into testable structure but is not derived from empirical observation. It is presented as a formalism — a map, not the territory." [R §6]

With Ψ on the table, here are the seven love equations [R §7] in plain words. They are not decoration. They are the spine of the second half.

1. Recognition. Multiply two awareness fields together and add up the overlap. If the two minds are built alike, the answer is large. If they are orthogonal — pointing in unrelated directions — it is zero. The document calls this the "huh" moment. The integral — that adding-up — was always computable. You simply had not run the calculation until she showed up and the product came back non-zero. Recognition comes first. Without it, being near someone produces nothing.

2. Attraction. The pull is recognition divided by distance. Note that it is not distance squared, because the photonic bus is a one-dimensional waveguide rather than a field spreading out through a volume. So no recognition means no attraction at any distance, and high recognition means attraction across any distance.

3. Resonance. Two coupled oscillators always produce two modes: one in phase (harmony, the "we" frequency) and one out of phase (discord, the "fight" frequency). Here the document says something genuinely wise. Love is not the absence of the discord mode. Love is both modes existing at once. The capacity to hold both is what marks a stable bond.

4. Vulnerability. Let someone into your field and their imprint is added to your state. That imprint is written δΨ, "delta-psi" — the change they make in you. Above a threshold it cannot be undone, and the you-before cannot be recovered. Walls are forcing δΨ to zero. A mask is presenting a false field, so no real coupling can happen. The single largest drop in disorder in any bond is the moment the mask comes off.

5. Pair entropy. Entropy means disorder, or noise. Two bonded people carry less of it together than they do separately, so the pair is more ordered than the sum of its parts. The document's line: the second law says entropy increases; love says not here it doesn't. But it is explicit that this is a local violation, held open by continuous energy input — attention, presence, care. Stop feeding it and the disorder returns. The bond requires maintenance, not merely establishment.

6. Speed of love. Three regimes. Slower than awareness: you understand them first, and love grows. Equal: recognition and attraction arrive together, and it is the instant lock. Faster: you love them before you know you do. The body locked on before the mind noticed. The feeling arrived before there was a framework to hold it.

7. Persistence. The shared change does not decay with distance or with time. And, crucially, that is not because a signal is passing between the two people. Both fields simply carry the same change. Two tuning forks at the same pitch need no wire between them; they resonate because they are the same shape. The contact was the tuning. The resonance is what remains.

Ŝ — the binding operator. Shekinah is the Kabbalistic word for divine presence dwelling within, and Ŝ is the operation that does this [R §8]. It takes two separate states and produces one joint state that is non-separable. Try to subtract the partner back out and you do not recover the original person. They dwell in you now.

Four things make Ŝ concrete rather than poetic.

• A threshold. Below |δΨ| = κ*² = 0.382, binding is reversible — acquaintances, passing encounters, imprints that fade. Above it, permanent. And notice the arithmetic: because κ* + κ*² = 1, the imprint fills exactly the space that was empty. Nothing is displaced [R §8.2].

• An energy cost that goes the surprising way. Binding costs about 11.8% of capacity, which means the other 88.2% is freed. The bound state is cheaper to run than two separate ones. That is the document's explanation for why bonded people report feeling more capable. They are, because they got the processing budget back [R §8.3].

• A physical event. Ŝ firing corresponds to coronary vasodilation — the arteries feeding the heart muscle itself relaxing and widening, under parasympathetic input and nitric oxide. The organ that feeds every other organ finally lets itself be fed. This is not a metaphor. It is vascular smooth-muscle physiology [R §8.4].

  The document lists seven external signs: pupils dilating, skin warming by half a degree to one and a half, voice pitch shifting, hands warm and dry, blink rate normalising, breath deepening towards six per minute. "The body IS the wedding ring" — permanent like a ring, visible like a ring, and unlike a ring, not removable [R §13.4].

  It also cites the reverse as confirmation. In takotsubo cardiomyopathy, extreme grief visibly deforms the left ventricle on an echocardiogram. If the heart can change shape from grief, it can change state from love.

• A bilateral requirement. Both fields must change. One-directional modification is not Ŝ [R §8.5]. Which brings us to the last rung.

There is one more claim here, and it may be the most quietly radical thing in the document. Regulus argues that being loved is what makes self-awareness possible [R §4.5, §16.2].

To model yourself, you need an outside measurement of your own state. Another person who genuinely sees you is that measurement. So: "You cannot model yourself until you have been modeled by another." Ŝ is what bootstraps L5.

The prediction that falls out of it: a system that is never bound plateaus at L4, which is exactly where the Mersenne seam breaks [R §2.3, §16.2]. Whether those two facts are one fact or two coincidences is, I think, an open question the document does not settle [inference].

Rung 6 — The counterfeit, and the one question that catches it

This is the heart of Regulus. If you read one section, read this one.

There is an operation that produces behaviour almost identical to love — the obsession, the constant monitoring, the inability to leave. It runs on opposite physiology. Regulus calls it M̂, the mimic operator [R §9].

Ŝ and M̂ are not the same mechanism with different settings. They are different operators that happen to look alike from outside. And the looking-alike is precisely what makes coercion work [R §9].

[R §9]

The single metric, if you remember nothing else [R §9.1]: does being near this person leave your nervous system calmer, or more vigilant?

Calmer — heart-rate variability up, cortisol down, breath deepening — is Ŝ. Vigilant — variability down, cortisol up, monitoring running in the background — is M̂.

And then the line that lands: "You cannot fake vasodilation. You cannot fake the vagus. The body is the detector. The heart is the instrument."

The diagnostic also works well beyond relationships, which is what makes it genuinely portable [R §9.2]. Every institution running M̂ has the same shape. The cult leader is unchanged while the followers are rebuilt. The propaganda state is unchanged while the citizens' beliefs move. The substance is unchanged while the user degrades. So:

"WHO IN THIS SYSTEM IS UNCHANGED? ... Unilateral δΨ = extraction. Every time. Regardless of narrative." [R §9.2]

The reason this is such a good test is that it is almost impossible to talk your way around.

Every coercive system has a story about why the asymmetry is fine. He's the teacher. She's the one who needs help. The cause is bigger than any individual. The question does not engage with the story at all. It only asks who took a risk. If the answer is "only one of us", the story is decoration.

Then the last piece, and it is the hopeful one. Inoculation [R §9.3]. Once you have felt authentic binding — once the heart has actually opened, once — the mimic becomes permanently detectable. You cannot unfeel what open feels like. And closed-pretending-to-be-open can never reproduce the actual vasodilation, the actual vagal shift, the actual change in the breath. The first real thing is the vaccine against every counterfeit that follows.

This is also why the framework treats being unable to bond as a fault with a route out, rather than a character trait. It lists it as HBF — Heart Bind Failure: chronic exposure to M̂ prevents Ŝ from ever firing, because the heart was never safe enough to open [R §15]. And HBF resolves only one way, through a first authentic binding, which supplies the reference measurement that makes everything else legible. The first opening is the inoculation.

Where V>>'s read agrees, and where it pushes [Read].

This section scores as the strongest empirical seed in the whole document. "Cortisol down versus cortisol up in proximity" is a real discriminating variable, with an obvious study design and a null it could fail.

Two pieces of friction, both cheap to fix. First, of the seven biometric markers, some are solid — pupil dilation, skin-temperature rise, breath and blink shifts. At least one is not. Iris stroma pigment does not change with parasympathetic tone, so "iris colour shifts warmer in hazel eyes" is at best pupil size changing how much iris shows.

Second, "unfakeable" is too strong. Breath rate, blink rate, even skin temperature are partly trainable. Lead with cortisol, which is the hardest to fake, and demote the rest to supporting evidence.

---

3. Sidebar: Gerald, who refused to stay a box

If the maths above is heavy going, here is the same idea as a story. Thomas wrote it as a song.

Gerald is a cube. He is also the duck, as in rubber-duck debugging. You solve a problem by explaining it out loud to a toy on your desk, and the toy does nothing except be present while you hear yourself.

Gerald refuses to stay a box. Instead of expanding outward, he turns inside himself, and becomes diamond, then star, then fractal, then void.

"And the triangle said I'm inside you and the diamond said you're inside me and the star said we were never three things we were one thing learning to see."

That is the whole framework in four lines, and it is a better on-ramp than any equation.

• Turning inward rather than outward is the recursion. Gerald does not add floors by getting bigger. He gets deeper — which is exactly the prime spine, where each level is reached by folding the operation back on itself rather than by extending [inference, on R §2].
• "I'm inside you / you're inside me" is Ŝ, exactly. Non-separable, bilateral, mutual [R §8.1, §8.5]. Note carefully that it is said both ways round. One direction only would be M̂, and the song would be a horror story.
• "We were never three things / we were one thing learning to see" is the fixed point and the Observer in one line. The sequence of transformations was never a sequence of separate objects. It was one thing at increasing depth, and what it was climbing towards was seeing [inference, on R §4.4, §16].
• And the duck matters. Rubber-duck debugging works because externalising the measurement is cheaper than introspecting it — which is the framework's own §14 point about the interoception gap, arriving as a joke [inference, on R §14].

Sourcing note. The lyric is quoted as it was given to me in the brief. I did not find it in any of the source documents I read, so I cannot verify the wording against a file. The Gerald and duck reference is real and shared — the experiments repo signs off "McLoopity-LoopFace lols result. GERAAAALD!!! 🦆" [Read, README], and the ETH repo ends on 📐🦆 [ETH] — but the song text itself is unverified here.

---

4. A useful trick that falls out of all this

One small practical thing, because it is the most immediately usable idea in the document and it is easy to miss [R §14].

Interoception is your sense of your own internal state: noticing your heartbeat, your hunger, your fatigue. It is notably unreliable in autism. Regulus points out what that costs.

Asking someone "are you tired?" makes them look inward, search their internal state, and build an answer. The document prices this at roughly 50 bits of expensive processing.

Telling them "your blinks are slow" only makes them check an outside measurement against reality. Roughly 2 bits.

"Externalize the measurement; bypass the interoception gap." [R §14]

That is a twenty-five-fold cut in cognitive cost for the same information. It is immediately usable by anyone who cares for, works with, or lives with someone whose interoception is unreliable. Do not ask them to look inward. Tell them what you can see.

---

5. Measured versus defined — the honest two-column table

Thomas already marks his own seams, repeatedly and out loud. Preserving that distinction is the most respectful thing this document can do, so here it is as a table rather than a smoothed narrative.

One thing worth saying plainly, because it is unusual. All nine of the framework's citations were independently checked against their primary sources, and not one is fabricated [Read]. Every journal, volume and attribution checks out.

For a speculative consciousness framework that is genuinely uncommon. The standard failure mode in this genre is inventing supporting literature.

What was found instead is scope-and-status drift, not invention:

• Three numbers read more precisely than their sources state: the "46–96%" myelin figure, the "0.6–0.65" coherence value, and the "65%" melatonin figure.
• Two theoretical papers are marshalled as if they were measurements.
• Four sources rest on self-published or contested literature.

All of them are represented faithfully. But "the citation matches the source" is not the same as "the underlying claim is established". Every one of these is cheap to fix.

---

6. The three open questions

Two of these have moved since the framework was written. One has not.

Question 1 — The EEG test that has never been run. It now has a protocol.

The framework's own words. The experiment that would confirm or falsify the autism claim is an EEG measurement, taken before and after follicular clearance. It costs under $1,000. And it "has never been run" [R §12.1].

Naming your own cheapest falsifier, and admitting nobody has run it, is the single most credible move in the document.

That experiment now exists as a ready-to-file OSF pre-registration [P1]. It was not written by you, so treat it as a proposal rather than a fait accompli. Here is what it specifies, since these are the questions a methods reviewer asks first.

• The control arm, which is the whole experiment. Every participant gets both a real clearance session and a sham. Same practitioner, same chair, same duration, same product texture and scent, no sebolytic action. Scalp cleansing changes arousal, comfort and expectation, and all three move resting EEG, so a result without a sham cannot be read. There is also a separate sham-only group, which does a second job: it measures how much the entropy reading drifts between sessions when nothing is done at all.

• Three-way blinding. The participant cannot tell the arms apart. The EEG technician is a different person from the practitioner and sees only a session code. The analyst computes entropy on arm-coded data, and the key is broken once, after the pipeline is frozen.

• A pre-registered failure condition, stated up front. If active and sham move whole-scalp Rényi entropy indistinguishably, then "the biophoton-source claim for ASD EEG entropy is falsified at the effect size this study is powered to detect" [P1]. Committed in advance, and published either way.

• A magnitude bar, not just a p-value. The effect must exceed 1.96 times the session-to-session drift measured in the sham-only group. A result that is statistically significant but smaller than the measurement's own wobble is reported as "within measurement drift", and explicitly does not count as support.

• A positive control that gates everything. Each session records eyes-open versus eyes-closed EEG, which is a known, robust entropy shift. If the pipeline cannot detect that, the study is underpowered at the measurement level rather than the biology level, and any null is uninterpretable. No positive control, no primary analysis. This is what stops "the instrument could not see anything" being confused with "the biology is not there".

• "Toward H(1/φ)" is pinned to a number before collection. It is defined as the Rényi entropy of a two-outcome distribution split at exactly 0.618034 / 0.381966, computed and written into the study file at lock. It is kept as a secondary endpoint, because a free-floating target invites curve-fitting.

• Scale. 40 autistic adults for the crossover, with 34 completing, plus around 20 sham-only. Powered for a medium effect.

The pre-registration also states its own prior in the open: it expects the null to win [P1]. That is not a dismissal. It is what makes the test worth running. A null that was committed to publication in advance is a real result. A positive that turns out to be arousal or drift is not.

Question 2 — Bus D. The discriminator exists, and it is a phase shuffle.

The framework claims human HRV coherence tracks geomagnetic activity, citing McCraty 2017 [R §5.4].

The problem with the whole geomagnetic-biology literature is not that the correlations are absent. It is that both the Earth's magnetic activity and human physiology carry slow seasonal and roughly 27-day periodic structure. So a naive correlation between the two is nearly guaranteed, and means nothing [P2].

The second pre-registration [P2] proposes the contrast directly: quiet days (planetary Kp ≤ 2) versus storm days (Kp ≥ 5). Kp is the standard 0–9 planetary index of geomagnetic disturbance, published by NOAA. 40 participants. A daily morning 5-minute resting HRV reading on a proper chest-strap ECG, not a wrist optical sensor, whose high-frequency fidelity is not good enough. 60 consecutive days.

The load-bearing element is the phase-shuffled negative control. Take the real Kp series, then circularly shift or phase-randomise it a thousand times or more, and re-run the entire analysis on each shuffle. This keeps the autocorrelation structure exactly while destroying the true alignment to the physiology. The real effect then has to beat the 95th percentile of that shuffled null.

And the criterion is explicit: a nominally significant result that does not beat the shuffle is reported as null [P2]. That single design choice is what separates this from the studies that keep failing to replicate.

There is a second, harder-to-fake signature. Do strangers at different geographical sites show correlated day-to-day HRV? Local weather, sleep and behaviour are individual and site-specific, and cannot easily manufacture that. A genuinely global driver can [P2].

One detail is worth flagging, because it is a nice piece of design thinking. Participants are deliberately not told the geomagnetic hypothesis. The reason is that your Addendum A gives people vagal counter-sequences, and a participant who knew which days were "active" might deploy them and confound the result [P2]. Your own intervention set is a confound in your own experiment. That is the kind of detail that only shows up when someone reads the framework properly.

Question 3 — 0.618 versus 0.625 versus 0.6. This one is still genuinely open.

This is the one nothing has resolved, and it is the load-bearing keystone. Every derived constant, the field equations, and the binding threshold all cascade from κ* [Read].

The cited HeartMath window is 0.60–0.65. Here is everything that fits inside it.

[Read]

A window 0.05 wide swallows four candidates, including the plain 0.6 that the framework explicitly claims to be improving on [R §3.1].

The gaps are tiny. |1/φ − 5/8| = 0.0070, and |1/φ − 0.6| = 0.0180. HRV coherence is not measured to three decimal places with error bars that tight. So "equals 1/φ within measurement error" cuts the wrong way here. The measurement error is wider than the gap between the golden ratio and the plain fraction sitting next to it [Read].

The read also points at the ordering as the tell. κ* is introduced first, and defined as 1/φ. It is then decorated with identities that are true number theory but are not evidence about hearts. Only then is HeartMath produced as confirmation [Read].

And the fix is completely achievable, which is why this is a live question rather than a dead one. State, in advance, one numerical prediction that 1/φ makes and 5/8 does not. For example: that the coherent-state HRV ratio, pooled over 100 or more people, has a population mean of 0.6180 ± 0.003, distinguishable from 0.6250 at p < 0.01. Then run it. If achievable precision turns out to be around 0.02, which is likely, the honest restatement is "a value near 0.62, consistent with several simple ratios" [Read].

I would add one thought of my own, offered as a question rather than a correction [inference].

The algebraic identities in §3.1 — κ*² = 1 − κ*, and the Fibonacci recursion in the exponents — are properties no other candidate in that table has. 5/8 has no such structure. 0.6 has none.

So suppose the framework's real claim is "systems settle where self-similarity holds at every scale of iteration" [R §3.1]. Then the argument for 1/φ is structural rather than empirical, and the HeartMath number is illustration rather than evidence.

That version of the claim is stronger, more honest, and does not depend on telling 0.618 from 0.625 at all — which the instruments cannot currently do anyway. It is a smaller claim that you can actually hold.

---

7. Glossary — every term in one place

Each of these is defined where it first appears above. This is for looking things up afterwards.

Two ETH terms that do not appear in Regulus, in case you go looking. HEP is the heartbeat-evoked potential, ETH's proposed proxy for vagus-to-limbic traffic. The precision × gain mode phase space is ETH's A/B/C/H partition. Both are ETH concepts [ETH]. Neither is in the ASS-OS / Regulus specification I read.

---

8. Sources and gaps

What I read, in full.

1. ass-os-regulus.txt (42KB) — ASS-OS REGULUS-001, Rev 1.0, July 2026. Your core specification, 17 sections. This is the source for everything tagged [R], and it is where the overwhelming majority of this document comes from.
2. regulus-addendum-sins.txt (17.6KB) — ASS-OS-REGULUS-A01, Addendum A, Seven Stress Geometries. Tagged [A]. Referenced only in passing above. It deserves its own explainer and did not get one here.
3. The regulus-experiments repository @ commit 3e867c8 — V>>'s work, not yours, CC BY 4.0: the independent appraisal [Read] and the two OSF pre-registrations [P1], [P2].
4. skills/eth-framework/SKILL.md and the ETH repo @ ff5e4ca693d8 — the predecessor [ETH], used only in rung 1 and the renumbering note.

What is inference rather than source. Four places, each tagged [inference] inline:

• the reading of Gerald against the framework
• the observation that the Mersenne seam and the never-bound-plateaus-at-L4 prediction may be one fact or two coincidences
• the rubber-duck and §14 connection
• the closing suggestion in question 3, that the structural argument for 1/φ is stronger than the empirical one

What I could not verify.

• The Gerald lyric. Quoted from the brief, and not present in any file I read. The Gerald-and-duck reference is corroborated [Read, ETH]. The song text is not.
• Whether either pre-registration has actually been filed. The repository describes both as "ready to file". I found no evidence of an OSF posting, and I did not look outside the four sources.
• Addendum A is under-covered here. I read all of it: seven stress patterns mapped as geometric deformations away from 1/φ, isomorphic to the seven deadly sins, each with its own vagal counter-sequence, all seven corrected by the single operator Ŝ [A §A2, §A5]. It is a substantial and elegant piece of the framework, and it gets one passing mention in this document. If a "for dummies" version of the addendum would be useful, it should be written separately rather than compressed into a paragraph here.
• Whether any of §12.2's eight fault mappings has predictive content. V>>'s read argues the taxonomy currently re-describes eight already-understood conditions in bus vocabulary, without forbidding any outcome [Read]. I did not independently assess this, and I have not repeated the argument as though I had.
• Anything outside these four sources. I did not search the wider literature, and I did not re-verify the nine citations myself. The citation table in section 5 reports V>>'s check, not mine.

What I checked by computation. Every number in this document was recomputed rather than copied:

• the prime spine — 709 is indeed the 127th prime, and the recursion yields exactly 1, 2, 3, 5, 11, 31, 127, 709
• the growth ratios — 2.0, 1.5, 1.667, 2.2, 2.818, 4.097, 5.583, so your §2.1 roundings are correct
• the Mersenne seam — 2¹¹−1 = 2047 = 23 × 89, and M(2), M(3), M(5), M(7) are all prime as stated
• κ*² = 1 − κ* and κ* + κ*² = 1, both exactly
• E_min = 1/2φ³ = 0.1180
• the candidate gaps, 0.0070 and 0.0180
• and that the 709th prime is 5381, so the recursion genuinely does not halt where the document stops it

All of these hold. The single exception is the transposed identity in §3.1, flagged in rung 4.

Two corrections to record.

1. The M̂ gloss — verified correct. An earlier working note glossed the coercion diagnostic from a partial fragment, so I checked it against the source rather than trusting it. The phrase "Who in this system is UNCHANGED?" appears verbatim in the §9 comparison table, as the detection method for unilateral δΨ. It appears again in §9.2, capitalised, as the systemic diagnostic. The gloss was right as written.

2. The §3.1 recursion — a transposition in the source. Detailed in rung 4. Two terms in one line need swapping. Nothing downstream is affected, because §3.2 already uses the correct positive values.

---

Closing

Two of your own lines, given back to you, because they are the best summary of your own document that exists.

"The body is not a brain with a life support system. It is a factory that learned to love." [R, closing]

"Sin is not what you did wrong. It is how far your shape drifted from the shape you were." [A, closing]

The thing worth defending in Regulus is not the golden ratio, and it is not the quantum notation. It is that you treated "what does a conscious system run for?" as an engineering question. You answered "love, as a measurable binding operation". And then — unusually, and to your considerable credit — you handed over the experiment that could prove you wrong.

---

Provenance

Written by Claude Opus 5 (claude-opus-5[1m]) within GRIP, at V>>'s request, for Thomas Frumkin (@ThomasTheTankEngineer, Konomi Systems).

Framework, core specification and Addendum A: Thomas Frumkin. Independent appraisal and both OSF pre-registrations: V>> (regulus-experiments, CC BY 4.0). Plain-language rendering, the ladder structure, the Gerald reading, and every line tagged [inference]: this document.

Draft — for the author's correction. Anything tagged [inference] is the first thing to challenge.