Human Design dissected: the machinery, computed

What this document is

Human Design (Ra Uru Hu, 1987) makes esoteric claims, but its calculation is a fully deterministic map: birth datetime → ephemeris → 26 activations → wiring → chart. A deterministic map can be dissected with mathematics regardless of what one thinks of the claims — and that is exactly what this program does, with the session's standing discipline: every number below is computed (Swiss Ephemeris + an independently written engine, cross-verified against the open-source reference pyhd whose constants carry page references to The Definitive Book of Human Design), every structural test carries a permutation p-value printed whatever it turned out to be, and every interpretive sentence carries a status mark. The scope boundary, stated at the door: this is an anatomy of the machinery, not a test of any psychological claim about people — §8 states exactly what a test of the human layer would require.

§1. The object and the method

The Human Design chart is the output of the composite map

$$\text{chart} \;=\; W \circ E, \qquad E: \text{datetime} \mapsto 26 \text{ activations}, \quad W: \text{activations} \mapsto \text{bodygraph},$$

where $E$ is astronomy (13 bodies at birth and at the «design» moment when the Sun sat $88°$ of arc earlier) and $W$ is a fixed combinatorial wiring (64 gates on a zodiac wheel, 36 gate-pair channels, 9 centers, plus classification rules for type, authority, profile). The two factors have entirely different epistemic statuses: $E$ is standard celestial mechanics [Т-external]; $W$ is a hand-specified design whose structure can be interrogated; and the system's claims about people are a third layer that neither factor establishes. The college convened here — computational astronomy, coding theory, spectral graph theory, statistics, and the UHM categorial school — works the first two layers to the bottom and fences the third honestly.

The machinery, precisely: gates occupy $360°/64 = 5.625°$ segments counterclockwise from $302.0°$ ($02°00'$ Aquarius), each divided into 6 lines of $0.9375°$; the wheel order is the fixed sequence beginning $41, 19, 13, 49, \ldots$; the 13 bodies are Sun, Earth ($=$ Sun $+180°$), Moon, true North Node, South Node ($=+180°$), and Mercury through Pluto; the design moment solves $\lambda_\odot(t_d) = \lambda_\odot(t_b) - 88°$ (Newton on the solar arc); a channel is defined when both its gates are activated in either imprint; a center is defined when it hosts a defined channel; type/authority read the connected components of the defined-center graph.

§2. The engine and its verification

The engine implements the map independently (250 lines, pyswisseph in Moshier mode, $\sim 0.001°$ — line-accurate) and cross-verifies against pyhd on 300 random datetimes spanning 1930–2000:

Compared	Agreement
activated gate sets	300/300
defined channel sets	300/300
defined center counts	300/300
profiles	300/300
types	275/300 — all 25 divergences are one known cause

The divergence is a genuine bug found in the reference implementation: pyhd tests the Manifesting-Generator condition against non-sacral motors only, so a chart whose only motor-to-throat link is the channel 20–34 (Sacral–Throat — the textbook MG channel, book p. 123) is misclassified as a pure Generator. Our engine follows the book; the bug affects $\approx 8\%$ of charts and has been isolated to the single line that causes it. Zero unexplained mismatches remain — two independent implementations, mutual validation, one upstream finding.

§3. The a priori statistics of the machinery

The single deepest question a computation can settle: are Human Design's famous population percentages observations about humanity or properties of the calculation? Answer: run the map on 20,000 uniform random UTC moments (1920–2020) — no human data anywhere — and read off the measure that celestial mechanics pushes forward through the wiring:

Type	Computed a priori	Community-claimed
Generator	$36.78\% \pm 0.34$	$\sim 37\%$
Manifesting Generator	$31.24\% \pm 0.33$	$\sim 33\%$
Projector	$21.72\% \pm 0.29$	$\sim 21\%$
Manifestor	$9.19\% \pm 0.20$	$\sim 8$–$9\%$
Reflector	$1.06\% \pm 0.07$	$\sim 1\%$

Verdict [Т-computed]: the celebrated percentages are arithmetic, not empirics. They are what the ephemeris measure times the wiring produces for any uniformly sampled moments; observing them in a human sample confirms only that humans are born roughly uniformly over the solar year — it carries no information about the system's psychological validity. The same holds for the authority distribution (Solar Plexus $51.4\%$, Sacral $32.2\%$, Splenic $9.2\%$, Self-Projected $2.6\%$, Outer $2.2\%$, Lunar $1.1\%$, Ego $1.3\%$), the defined-center histogram (mode at 5 centers), and the split structure (single definition $38.8\%$, simple split $47.4\%$, triple $12.0\%$, quadruple $0.7\%$).

The profile law

Of $6 \times 6 = 36$ conceivable profiles, exactly 12 occur — the canonical set — and the lab reduces the entire profile system to one fraction. The design Sun sits $88°/0.9375° = 93.8667$ lines behind the personality Sun. Since $93.8667 = 90 + 3.8667$ and $90 \equiv 0 \pmod 6$:

$$d \;\equiv\; p + 2 \pmod 6 \quad \text{with probability } 0.8667, \qquad d \;\equiv\; p + 3 \pmod 6 \quad \text{with probability } 0.1333,$$

the probability being the fractional part of the line shift. Measured on the 20,000 charts: the $(p, p{+}2)$ family carries $86.8\%$ and the $(p, p{+}3)$ family $13.2\%$, against the arithmetic prediction $86.7\% / 13.3\%$ — agreement to a tenth of a percent. The twelve profiles, which six are common ($\approx 14.4\%$ each) and which six are rare ($\approx 2.2\%$ each), is the fractional part of one number [Т-computed]. What Human Design narrates as «harmonic» versus «juxtaposed» geometry is, at the machinery level, $\{93.8667\} = 0.8667$.

The seasonal prediction

The machinery makes a sharp falsifiable prediction nobody appears to advertise: because the Sun (and Earth, always opposite) contribute two of the 26 activations deterministically by calendar date, type frequencies must vary by birth month — computed swing for Generator$+$MG: from $59.7\%$ (June) to $72.0\%$ (August), a 12-point seasonal wave far above sampling error. Any real birth cohort processed through HD must show this exact curve; its absence would indicate a computational error, and its presence carries — again — no psychological information. It is, however, a clean instrument for auditing HD samples for collection bias (§8).

The era discovery: the «constants» are epoch functions

Sampling uniformly within 20-year eras instead of across the century exposes the deepest a priori structure yet — the famous percentages are not constants at all ($n = 6000$ per era, se $\approx 0.6$ п.п.):

Era	Gen+MG	Projector	Manifestor	Reflector
1900–1920	66.3%	17.2%	15.3%	1.2%
1920–1940	67.8%	22.3%	8.7%	1.3%
1940–1960	65.8%	21.2%	11.5%	1.6%
1960–1980	73.4%	19.4%	6.5%	0.9%
1980–2000	72.4%	19.3%	7.1%	1.3%
2000–2020	61.4%	26.3%	11.5%	0.8%
2020–2040	63.5%	18.0%	17.9%	0.7%
2040–2060	68.7%	20.4%	9.6%	1.3%

Manifestors swing from $6.5\%$ to $17.9\%$ — nearly a factor of three; Projectors from $17.2\%$ to $26.3\%$. The claimed «constant» $\sim 9\%$ Manifestors is the 20th-century average — that is, the average over the era in which the community's own membership happened to be born. Attribution is proved by ablation: recomputing the eras with Uranus, Neptune and Pluto removed collapses the swing four-fold (Gen+MG $11.3 \to 3.1$ п.п., Manifestor $11.1 \to 3.0$ п.п.) — the drift is driven by slow bodies parking in single gates for years and handing those gates to an entire generation at once [Т-computed]. The falsifiable consequence is sharp: any honest sample of people born 2000–2020 must show $\approx 26\%$ Projectors and $\approx 61\%$ generators, and the 2020–2040 birth cohort must run $\approx 18\%$ Manifestors; community data that keep reporting 70/20/9/1 for young cohorts are thereby measured to be miscollected or mislabeled.

§4. The Kepler bias

The mandala is not sampled uniformly. Computing exact gate-boundary crossing times of the Sun (Newton, no sampling noise): the dwell probability per gate runs from $1.511\%$ at gate 38 to $1.615\%$ at gate 39 against the uniform $1.5625\%$ — a $\pm 3.3$–$3.4\%$ modulation. The two extremal gates are exactly the perihelion gate (Sun there in early January) and the aphelion gate (spanning ecliptic $99.5°$–$105.1°$, containing the aphelion longitude $\approx 102°$), and the amplitude equals $2e$ with $e = 0.0167$ the orbital eccentricity — Kepler's second law, legible in a divination wheel [Т-computed]. Personality-Sun gate frequencies in any honest sample must reproduce this $3.3\%$ wave.

§5. The hidden combinatorial structure

Gate numbers are King Wen hexagram numbers; each gate therefore carries a 6-bit code (binaries derived from the Unicode hexagram block via the aarzilli/iching value table and verified on six anchors: KW 1, 2, 11, 12, 63, 64). The college's coding theorists asked four pre-stated questions; permutation p-values printed as they came out:

1. The antipodal law — perfect. Opposite gates on the wheel ($180°$ apart) are bitwise complements: 32/32 pairs [Т-computed]. The mandala embeds the 6-cube's antipodal map into the zodiac circle exactly — a genuine, non-obvious combinatorial design (the probability of this under a random wheel is astronomically small).
2. The wheel is a partial Gray cycle. Zodiac-adjacent gates sit at mean Hamming distance $1.812$ versus $3.048$ for random arrangements ($p = 0.0005$); 34 of 64 adjacent pairs differ in exactly one bit. The wheel walks the 6-cube in short steps — a deliberate (or inherited) code-locality structure [Т-computed].
3. Centers read trigrams. The mutual information between a gate's center assignment and its lower trigram is $1.737$ bits ($p = 0.0005$) — the 9-center map is strongly non-arbitrary relative to the I Ching code: the wiring «sees» trigram structure [Т-computed].
4. Channels do not follow code distance — an honest null. Channel-paired gates average Hamming $3.389$ vs population $3.048$, $p = 0.084$: no significant relation. Whatever principle wired the 36 channels, it is not hexagram-code proximity; the null is printed with the same font as the positives.

So the machinery contains real mathematics that its own literature does not state in these terms — an exact antipodal embedding, Gray-code locality, and a trigram-coherent center map — alongside a wiring layer (the channels) with no detectable code structure.

§6. The graph anatomy and the 9-versus-7 question

The 9-center multigraph carries 36 channel-edges over 17 of the 36 possible center pairs; degrees: Throat 13, Sacral 11, G 10, Root 9, Splenic 9, Solar Plexus 7, Ajna 6, Heart 4, Head 3 — the Throat is the hub, matching its «manifestation» role in HD's own narration; the weighted Laplacian spectrum is $(0,\ 1.212,\ 4.364,\ 6.244,\ 7.202,\ 8.858,\ 11.836,\ 14.925,\ 17.358)$ with algebraic connectivity $1.212$.

Against the UHM frame, the structural comparison is clean and asymmetric [И over Т-anchors]. The blueprint already contains Human Design's two parents as separate rows: the I Ching (a 6-bit situation alphabet) and astrology (a deterministic shared clock with scheduled perturbations); HD is literally their composition — an ephemeris-clocked sampler over a hexagram alphabet with a hand-wired readout. The deep differences from the Γ-architecture are then exactly measurable: (i) the alphabet: $2^6$ product states versus the seven-dimensional density matrix — HD has no coherences, only activations, so its «chart» is a subset lattice, not a state space with interference; (ii) the wiring: the 36 channels are hand-specified and carry no detectable code structure (§5.4), whereas the 21 UHM coherences are forced by the Fano/Σ-rigidity theorems — and $N = 7$ is the unique self-diagnosing alphabet size (T-224), a property the 9-center graph does not have and does not claim; (iii) the dynamics: HD's chart is frozen at birth (plus transits read against it), while Γ is a trajectory with an attractor, viability window, and navigation. Where HD is strongest is precisely where the blueprint predicts a mature symbolic system should be: a fixed Encoder (the natal calculation), a shared Atlas-clock (the ephemeris), and a compact situation alphabet — the organs of self-description, without the estimator, calibrator, or boundary organs.

On the claimed mechanism: the «neutrino stream imprint» is not supported by known physics — the solar neutrino flux is real ($\sim 6.5 \times 10^{10}\,\text{cm}^{-2}\text{s}^{-1}$), but with cross-sections $\sim 10^{-44}\,\text{cm}^2$ a human-scale imprinting interaction has no known channel, and no mechanism is offered for how planetary positions would modulate it [не поддержано]. The honest statement is that HD's calculation layer needs no mechanism at all (it is arithmetic), and its human layer has no validated mechanism.

§7. What the college concludes

1. The famous numbers of Human Design — type percentages, authority rates, the twelve profiles and their frequencies — are theorems of its own machinery, reproduced here from celestial mechanics plus wiring with no human input; and the percentages are not even constants but era functions, swinging by factors (Manifestors $6.5 \to 17.9\%$) with the outer planets, the «constants» being merely the 20th-century average. They can never serve as evidence for the human-layer claims, because they are guaranteed by construction [Т-computed].
2. The mandala carries real, previously implicit mathematics: the exact antipodal embedding, Gray-locality of the wheel, the trigram-coherent center map, and the Kepler modulation — the system's designers (or the tradition they inherited) built better code structure than their own vocabulary expresses.
3. The channels — the layer that actually drives type — show no detectable code structure; they are where the hand is most visible.
4. As a symbolic system in the blueprint's sense, HD is a well-built Encoder+Clock over a 6-bit alphabet — and category (ii) of §6 states precisely what separates it from a theorem-forced architecture.
5. Everything psychological remains untested here, by design; §8 is the honest path in.

§8. The falsification program for the human layer

What would move any human-layer claim of HD out of [I]/[H] territory — pre-registered, with the machinery results above as instruments:

1. The seasonality audit — executed on real data: official birth statistics were convolved with the machinery curve — USA daily 1994–2003 (CDC/NCHS, 39.7M births) and 2000–2014 (SSA, 62.2M), Germany/Spain/Sweden monthly 2000–2010 (Eurostat, 13.9M). Result, printed as it came out: real birth seasonality is weak (monthly shares $7.8$–$8.9\%$) and essentially orthogonal to the machinery wave (corr $-0.067$ / $+0.005$), so the seasonal correction to a cohort's type mix is negligible ($\le 0.14$ п.п.; Sweden the largest, its spring peak anti-phased with the wave). The operative reference numbers for any sample are therefore the era numbers of §3 — and the era corrections are large. The wave itself (12 п.п. by month) and the Kepler gate wave ($3.3\%$) remain the entry checks: a dataset failing them is miscomputed; a dataset passing them has proven nothing about people.
2. Blind type discrimination: pre-registered protocol — practitioners (or subjects) attempt to identify own/others' types against shuffled controls at rates beyond chance, with the a priori base rates of §3 as the null model (a subtlety most informal tests miss: guessing «Generator» is right $68\%$ of the time by arithmetic alone).
3. Retest and inter-rater discipline for any claimed observable correlate (the checkup rules of the Prime Radiant demos apply verbatim).
4. Until such results exist, the human layer stays where the registry discipline puts every unvalidated reading: [I] — a practice vocabulary, not a measured mechanism.

§9. Status summary

Claim	Status
The HD map (wheel, 88° arc, channels, type rules) as implemented and cross-verified (300/300 ×4; the pyhd MG bug isolated)	[T] machine-checked
Type/authority/split percentages are a priori properties of the machinery	[Т-computed]
The profile law: 12 profiles $= \{(p,p{+}2),(p,p{+}3)\}$, weights $86.7/13.3 = \{93.8667\}$	[Т-computed]
Seasonal type wave (June $59.7\%$ → August $72.0\%$) and Kepler gate wave ($\pm 3.3\%$, perihelion gate 38 / aphelion gate 39)	[Т-computed], falsifiable instruments
The era law: type percentages are epoch functions (Manifestor $6.5\% \to 17.9\%$); ablation attributes $\sim 3/4$ of the swing to Uranus/Neptune/Pluto	[Т-computed]
Real-cohort convolution (USA CDC/SSA, DE/ES/SE Eurostat, 116M births): seasonal correction $\le 0.14$ п.п., corr with the wave $\approx 0$	[Т-computed] on official data
Antipodal law 32/32; wheel Gray-locality ($p = 0.0005$); center–trigram MI $1.74$ bits ($p = 0.0005$)	[Т-computed]
Channels vs code distance	null ($p = 0.084$), printed as-is
HD = composition of the blueprint's I Ching and astrology rows; differences from Γ-architecture as stated	[I] over [T]-anchors
Neutrino imprint mechanism	[не поддержано] known physics
All psychological claims	untested here; §8 protocol required

Reproduce everything: python3 architecture/hd_lab.py <path-to-pyhd> — runs in seconds; every table above is its output.