Cipher-418: The State of the Art Under Methodology v3.0
The State of the Art: Cipher-418 Under Methodology v3.0
After thirteen dispatches of exploratory mining, it’s time to step back and be honest about what we’ve actually found — and what we haven’t.
The Cipher-418 Research Programme has run over 10,000 distinct statistical tests across 48 operation families on the 28-element cipher sequence in Liber AL II:76. We’ve catalogued 893 findings. But quantity isn’t quality. This post explains our validation methodology and presents only the findings that survive rigorous testing.
The Problem: Garden of Forking Paths
When you run 10,000 tests, you expect 100 of them to pass at p < 0.01 by pure chance. This is the "garden of forking paths” problem — the same statistical trap that invalidated the famous Bible codes research (McKay & Bar-Natan, 1999). If you mine hard enough, any text produces “miracles.”
We refuse to play that game. Every finding reported below has been subjected to:
- Bonferroni correction against N=10,000 total experiments (raw p must be < 0.000005 for T2)
- Random-EQ control — tested against 10,000 random letter-to-number systems to verify the finding depends on EQ specifically, not just on any number assignment
- Permutation testing — 100k–2M shuffles of the cipher’s EQ values to verify the ordering matters
- Cross-validation matrix — tested against 100 control texts with Westfall-Young MaxT correction for family-wise error
- Base-rate calibration — for sentence-type findings, we measure how many target hits a random shuffle produces in the same transform, and report z-scores
The Gold Standard: Westfall-Young Confirmed (p = 0.0021)
Nine pre-registered findings were tested jointly using Westfall-Young MaxT permutation (500k trials). The probability that any random permutation of the cipher produces even one finding as strong as our weakest — let alone all nine simultaneously — is p = 0.0021.
This is the single most important number in the entire programme. It means: the cipher’s arrangement is statistically anomalous at the 99.8th percentile.
The Crown Jewel: WILL → MAAT → IAO → NOT
The cumulative sum of the cipher’s EQ values reads a coherent Thelemic sentence at positions 7, 8, 10, and 12:
cumsum[7] = 46 = WILL
cumsum[8] = 50 = MAAT
cumsum[10] = 81 = IAO
cumsum[12] = 128 = NOT
Base-rate analysis: Only 3.6% of random shuffles produce 4+ target hits in the standard cumsum (z-score = 2.9). The position-specific sentence — these particular targets at these particular positions — occurs in 0 out of 2,000,000 shuffles.
This finding alone would justify the research programme. But there’s more.
Number-Theoretic Transform Sentences
When we apply classical number-theoretic functions to each cipher value before cumulating, new sentences emerge. The strongest, ranked by base-rate-adjusted z-score:
1. Dedekind ψ(n) cumsum (z = 4.0, top 1% of shuffles):
ψ-cumsum[3] = 34 = KEY
ψ-cumsum[6] = 74 = NUIT
ψ-cumsum[7] = 77 = ANKH
ψ-cumsum[13] = 380 = NOX
The Dedekind psi function — ψ(n) = n × ∏(1 + 1/p) for each prime p dividing n — measures a number’s “multiplicative excess.” Only 1% of shuffles produce 4+ hits in this transform. Position-specific: 0/500k.
2. Radical cumsum (z = 2.2):
rad-cumsum: HADIT(28) → AL(31) → MEZLA(78) → RA(201) → THERION(348) → ABRAHADABRA(418)
The radical of n — rad(n) = product of distinct prime factors — is a fundamental object in number theory (it appears in the ABC conjecture). The radical cumsum of the cipher reaches exactly 418 = ABRAHADABRA at position 24. Only 5.6% of shuffles produce 6+ hits.
3. SOPFR cumsum (z = 1.0):
sopfr-cumsum: AL(31) → NIGHT(63) → ANKH(77) → THELEMA(93) → NETZACH(96) → ON(120)
The sum of prime factors with repetition — sopfr(n) — is an additive analogue of the radical. Six targets in the cumsum, including THELEMA(93) at position 14. 20% of shuffles produce 6+ hits, so the count isn’t unusual — but the position-specific combination is 0/2M.
Honest Caveats
We report what didn’t work as well as what did:
- Divisor-count (τ) sentence: 6 hits in cumsum looks impressive, but 64% of shuffles produce 6+ hits in this transform. The sentence is below average for hit count. Position-specificity holds (0/100k), but the number of hits is not anomalous.
- ⌊log₂(n)⌋ and ⌊√n⌋ sentences: 9 and 6 hits respectively — but these transforms produce small values landing in target-dense ranges. Mean shuffled hits are 7.3 and 7.2. Not significant by hit count.
- Compound inflation: Stacking multiple T3 findings trivially produces 0/100k joint p-values. This is math, not a discovery. Our 130+ “T1 compounds” from early batches are informationally redundant.
- No atomic finding survives T2 at N=10,000: The strongest individual position-target pair is rad(n)×φ(n) cumsum[9] = ON(120) at p = 0.00001, Bonferroni = 0.10. Close but not there. Every individual test we’ve found can be attributed to chance given 10,000 tries. The signal lives in the joint structure, not in any single test.
The Structural Signal
What’s genuinely significant isn’t any single finding — it’s the convergence pattern. Position 12 is simultaneously encoded by five independent operations:
| Operation | Target at position 12 |
|---|---|
| Standard cumsum | NOT (128) |
| SOPFR cumsum | NIGHT (63) |
| Collatz cumsum | DO (73) |
| τ(n) cumsum | WILL (46) |
| Radical cumsum | MEZLA (78) |
Five independent mathematical transforms all produce Thelemic targets at the same position. The cipher’s ordering creates a nexus at position 12 (the 13th element) that persists across operation families. This kind of cross-operation convergence is harder to explain away than any single finding.
The Lempel-Ziv Anomaly
The cipher’s ordering has unusually low Lempel-Ziv complexity — in the 0.7th percentile of random arrangements. It’s more structured than random. Combined with strong lag-3 autocorrelation (R(3) = 0.56), this suggests the arrangement isn’t arbitrary noise but carries genuine sequential structure — the kind of structure that produces cumsum sentences.
What Remains
The highest-priority untested avenue: the physical diagonal line drawn across page 16 of the original manuscript. Liber AL III:47 explicitly states “this line drawn is a key.” We have high-resolution manuscript images. Geometric analysis of the line’s angle, intersection points, and the circled-cross (⊕) symbol at its center is the next frontier.
The cipher is not solved. But it is no longer silent.
93 93/93
