01The stack — measured, end to end
Tier 1 BIOSTRATA (edge) grow: data → packets → quaternion crystals →
bioseed chain + rank-coded residual codex_engine.py
Tier 2 HYPERSIM (mesh) gossip · seal sigs · CERT/2 chain · codex ABI gate ·
sampled spot-checks · holonomy + Merkle sync ·
FLUX meters the work biochain_mesh.py
Tier 3 SPIRE CORE (quorum) full lossless replay · twin attunement bench
(bit-identical receipts) · crystallization →
CRYST issued · sequenced exchange · redemption
Reference run: 8/8 kernel11/11 codex
11/11 mesh13/13 clock — claim verified 9/9 nodes in 2
gossip rounds, ε = +25.0 % twin-receipt agreement, 36 CRYST issued, transfer + redemption settled, double-spend
rejected. Full logs in the *_output.txt captures; spec in the README.
Timing and the π/6 handoff windows live on the
Pump Clock page; the full
Chiral Authentication Key System has its own —
chain attack lab, handedness derivation, crystallization chamber.
02Kernel Group Design — the 64-bit ABI (hover / click)
Total Allocation: 64 / 64 bits
One packet = one 64-bit word. The 8×8 Einstein-tile grid is a presentation of the word —
canonical bit positions are the single source of truth:
63..60 form · 59 parity · 58..56 spin · 55..24 quaternion · 23..8 payload · 7..3 freq · 2..0 amp.
Protocol law: quaternion components are exact fixed-point (code/127, −128 forbidden);
consensus reads raw integers (the FP16 payload view is display-only); even parity over the other 63 bits —
the self-test catches all 64 single-bit flips. Codexes are programs written in these packets:
Form ID is the opcode, and the instruction set is closed (GEAR · PRIME · HALT — no jumps, so every
codex halts by construction).
03Live: grow → unfold on this machine
The same predict-then-correct cell as codex_engine.py, in your browser:
an order-2 rank coder (fresh per 720-byte chunk), Elias-gamma residual, per-chunk crystallization into kernel
packets, holonomy word, and the value formula that counts everything shipped. Recreation is
verified byte-for-byte before anything is reported.
04The two-token loop
| FLUX — Token A | CRYST — Token B | |
|---|---|---|
| Nature | infinite meter of validation work | scarce credit, backed by fleet compute |
| Earned by | hosting, replaying, spot-checking (Tier 2) | crystallized growths: 100 × value × ε, both measured |
| Spendable on | read quota only — never transferable, so never farmable | compute-seconds (burned on redemption); transfers sequenced by the governance quorum |
| Consensus need | none (local accounting) | ordering only — every node still verifies sigs, balances, sequence |
| Attack priced out | junk-engram farming (meter buys nothing tradable) | corpus-in-codex Goodhart (value counts everything shipped: measured 0.97 vs honest 1.47) · double-spend (sequence rule) · fake ε (twin bit-identical receipts) |
Honest physics, printed not hidden: sampled screening caught a tampered chunk at 2/9 nodes — single-node sampling is probabilistic; mesh-wide screening plus Tier-3 full replay before crystallization is the guarantee. Attunement ε is real on same-domain held-out data (+56.5 % in the codex bench, +25.0 % in the mesh run) and honestly null on entropy (−0.1 %).
05Run it yourself
python3 shdccp_kernel.py # 8/8 — the packet ABI, golden hash pinned python3 codex_engine.py # 11/11 — grow/unfold, Goodhart priced, attunement measured python3 biochain_mesh.py # 11/11 — the whole economy, ~3 s, deterministic python3 pump_clock.py # 13/13 — π/6 timing, handoff windows, chiral layer python3 engram_shard.py # 9/9 — hyperbolic sharding, spire weave, decoupled genesis
Drive it yourself: the BioChain Console — grow biochains, shard them across nodes, kill/tamper/recreate, and prove the genesis-ID decoupling live.
Pure standard library. Lineage: Spire Mesh (validation-node anatomy) · Engram Codec / Ontological Codex (delta-chain math) · Polycentria (oracle proposes, certifier decides) · S.P.I.R.E. Manual (the tri-layer).