Owl Academy | The Chiral Authentication Key System

01Why chirality — the three centralization pressure points

Everything in the mesh already verifies signatures; chirality decides who may sign what. The design attacks the three places where the system could quietly re-centralize:

pressure point	without chirality	chiral answer
The root of trust	one master-key list — whoever holds it mints all authority	dual chiral anchors: genesis bakes an ⟲L party and an ⟳R party; every certificate and crystallization needs both hands
Self-dealing	a grower could attest its own claims through sock-puppet approvals	pairing rule: growth by one hand may only be attested by the opposite hand — your hand is fixed at sealing, so you can never approve yourself
Chain history	the XOR holonomy is commutative — reordering chunks is invisible (measured)	chiral holonomy: the ordered quaternion product + its mirror; order is part of the fingerprint

One law above all of it: chirality is structure, not hardness. ECDSA seals remain the only lock. Quaternion math is trivially computable — it partitions authority; it never replaces cryptography.

02Chain attack lab — XOR vs the chiral holonomy, live

A bioseed chain of 8 chunks, each crystallized into a kernel packet (the real fold, in your browser). The chain commits three fingerprints: the commutative XOR word, the ordered ⟲ chiral holonomy, and its ⟳ mirror (conjugates, reverse order). Run the attacks:

Reference numbers from pump_clock.py §5: forward vs mirror differ by 2.86 rad (non-commutativity is real) · swap 2↔5 → XOR word bit-identical (blind) while the chiral holonomy moves 1.77 rad (caught) · content tampering → both catch. Every claim should therefore ship {xor, chiral⟲, mirror⟳} — three words, 40 bytes, and reorder/splice/time-reversal attacks all become visible.

03Handedness — derived, never chosen

A node's hand comes from the digest parity of its sealed Cosmological ID — the registrar is write-once, so after sealing the hand is public, verifiable by anyone, and unpickable. Type any ledger name to walk the derivation (same rule as pump_clock.py handedness()):

…

Digest parity splits the network ~50/50 at scale (measured 4L/6R at n=10). Grinding fresh IDs to pick a hand is cheap — and buys nothing: hands carry no authority without tier certificates, which chain to the dual anchors. Quorum thresholds should follow the live hand census.

04The crystallization chamber — pairing rule + dual anchors

A claim crystallizes only when the chamber is chirally balanced: the grower's hand on one side, ≥2 opposite-hand attesters on the other, under an ⟲L + ⟳R anchor pair. Pick a grower and toggle attesters/anchors — the chamber shows why it passes or fails (same logic as pump_clock.py §6, 4/4 checks).

grower (one)

attesters (toggle)

anchors (toggle)

05Attack gallery — what dies, what survives (honest)

Single-root capture

impossible by structure

The genesis snapshot with one-handed anchors is rejected by every validator (measured). Capture now requires compromising two independent parties — and tick-12 cross-rotation keeps the pair moving.

Self-attestation

impossible by type

Your hand is fixed when your Cosmological ID seals. You can never be your own opposite hand, so grow-and-approve sock-puppetry fails at the type level, not the policy level.

Same-hand collusion ring

rejected

An L-grower with L-friends has zero valid attesters, however many friends it has. A ring must recruit across the chiral divide — which is exactly the population it can't pre-select.

Chain reorder / splice

caught — Δ 1.77 rad

Invisible to the XOR word (measured), visible to the ordered product. The mirror commitment doubles the constraint: a forger must satisfy both orderings simultaneously.

Sybil identity farming

unchanged — be honest

Hands don't stop Sybils. Grinding IDs for a hand is cheap; but hands carry no authority without certificates, so grinding buys nothing. Sybil resistance still lives in the cert chain.

Two-anchor collusion

residual 2-party risk

The L and R anchors can still conspire. Mitigations: tick-12 rotation, public chronicles, and the fact that every validator checks the pair independently. Defense-in-depth, not a proof.

Quaternion ops as crypto

never

The chiral layer decides who may sign what — never whether signatures can be forged. ECDSA P-256 seals remain the single cryptographic lock in the whole system.

Hand-population imbalance

monitor

Digest parity trends 50/50 but small networks skew (4L/6R at n=10). Attestation thresholds should scale with the live census so neither hand becomes a bottleneck or a chokepoint.

06Where it lives in the stack

grow      Biostrata edge   claim ships {xor, chiral⟲, mirror⟳} + grower seal (hand = L|R)
validate  Hypersim mesh    every node: ECDSA sig → cert chain → codex ABI → spot-check
                           → BOTH holonomies replayed → hand-typing of all signers
attest    opposite hand    ≥2 attesters whose hand ≠ grower's — enforced by type
crystallize SPIRE Core     at tick 6/12 windows · dual ⟲L+⟳R anchor countersign
rotate    every tick 12    L-anchor signs R's successor, R signs L's — the pair moves

Reproduce every number: python3 pump_clock.py (13/13 — §5 chiral holonomy, §6 pairing rule & anchors). Related: Pump Clock & π/6 handoff windows · Enterprise console · spec.