Grok Path verified - and some new findings the old research missed

What changed

A chapter "Grok Path Verification" was published in a sidebar section called "CFB Signature Analysis". It was set aside in September 2026.

The short version

On 2026-01-13 CFB posted 40 numbers in a public exchange with the Grok AI. An older internal research note (2026-01-14) claimed those numbers were 20 coordinate pairs navigating through the Anna Matrix, and that the sequence was anchored to two of his 2014 Bitcoin vanity addresses. We went through every claim with scripts and reported the outcome with explicit confidence tiers.

Tier 1 - as originally labelled (the "New" items stay open, see the re-check note above)

• All 20 matrix values at the claimed coordinates match exactly (20 / 20, zero errors).
• The 1CFB vanity address hash160 ends with bytes [72, 2, 45]. Byte 19 = 45 = row of path step 1.
• The 1CFi vanity address hash160 has bytes [82, 39, 145] at positions 17, 18, 19. Bytes 17 and 18 = coordinates of path step 20.
• New: Start row (45) + end row (82) = 127 --- the Mersenne mirror axis. The path begins and ends on matrix-mirror rows.
• New: Coordinate sum of step 1 = 137 = inverse of the fine-structure constant (1/alpha). The earlier research tried to argue this via a stretched interpretation of the matrix value (-118 unsigned = 138, off by one). The cleaner observation is that the coordinate sum is exactly 137.
• New: Byte 17 of 1CFB = 72 = PUSHBYTES_72, the same opcode documented in a Genesis chapter that has since been removed. And step 3 of the path has coordinate sum 14 + 58 = 72. The number 72 appears across at least four independent structural layers now.
• New: 1CFB byte 17 - byte 19 = 72 - 45 = 27, the value of a BTC address selection that was later not confirmed (off-by-one index error).

Tier 2 - observations with weight but not proof

• The path's values are richer in "meaningful constants" than random (6/20 hits vs 3/20 random mean), but the z-score is only 1.86 --- borderline, not significant at 95%.
• The path visits 3 known-special rows (Row 6 Oracle, Row 14 Bridge, Row 72 Genesis) out of 20 steps. Random expectation would be about 0.5.
• The second half of the path (steps 11-20) has rhythmic +10 to +13 row/column deltas broken by -34 and -38 jumps. The first half looks random. This suggests a two-phase construction with a deterministic rule in the second half.

Tier 3 - had to downgrade

• The "Master Key" claim: Two earlier source documents claim a 20-character polyalphabetic key derived from shifts along the path. They disagree with each other on the shift count (20 vs 19) and neither source computes its own claimed key correctly. Downgraded to speculation pending access to the Qubic identity database.
• "1 in 16 million" probability: Valid only under pre-registration. Post-hoc selection across 20 byte positions per hash160 and the full coordinate space weakens this to roughly 1 in 40,000. Still non-trivial, but not cosmically rare.
• "Seed IDs" like 10641, 7922, 1472 attached to each step: unverifiable without external data.
• "Satoshi Troll" 133 = 7*19 claim: upgraded with proper source. We traced the key back to bitcointalk.org thread 506692 (2014-03-07), not ning.com as local notes said. The arithmetic is 100% verified (2432 = 2^7 x 19, 2299 = 11^2 x 19, diff = 7 x 19) --- now Tier 1. The original poster self-identifies as a troll and explicitly says they are not Satoshi; whether the math is coincidence, grinding, or a same-author signature remains Tier 3.

Why we kept the speculative parts

Transparency. When earlier research has made stronger claims than the evidence supports, the right move is not to delete the claims --- it is to separate them from the verified facts and label them clearly. Readers can then judge the interpretive layer on its own merits.

What this unlocks

Earlier text called the PUSHBYTES_72 connection in 1CFB byte 17 a new Tier 1 finding; that reading stays open. The path visits rows that hit a previously documented "special row" set more often than chance would predict. And the second half of the path looks generative. If we can figure out what rule produces steps 11 to 20 from step 10, we may be able to extend the path and test whether it hits further anchor addresses.

The chapter was set aside in September 2026. See September 2026 re-check.