Arrekta: what to reuse, and what makes it different

Research note, 2026-10-18. Companion to PRODUCTION_PLAN.md.

Two questions:

How much existing open-source chain code can we extract so we move faster?
What makes Arrekta worth using when Bitcoin, Ethereum, Solana and the rest already exist?

Sources are listed at the end. Most coverage of PQ chains is secondary or vendor-written. Claims marked (verify) should be checked against primary docs before we use them publicly.

1. The landscape (October 2026)

The big networks are not quantum-safe yet, and won't be for years

Chain	Status	Timeline
Bitcoin	BIP-360 (P2MR: keep public keys off-chain until spend) and BIP-361 (sunset of legacy signatures) are drafts. ML-DSA opcodes exist only on a third-party testnet (BTQ "Bitcoin Quantum" v0.3.0).	A BIP-360 co-author estimates ~7 years. Other estimates say 5–10. Millions of BTC sit in addresses with exposed keys.
Ethereum	Feb 2026 PQ roadmap (pq.ethereum.org), hash-based validator signatures ("Lean Ethereum"). EIP-8141 frame transactions are under consideration. There is no efficient ML-DSA precompile yet.	Core milestones around 2029.
Solana and most other L1s	No PQ signatures on mainnet.	—

The world's largest asset ledgers need half a decade of coordinated migration. Meanwhile, regulators are publishing PQC deadlines; Canada's mandate is already in force (verify scope).

Chains that already claim quantum resistance

Chain	What they did	Gap
QRL	PQ since 2018 with XMSS (stateful hash-based, so key reuse is dangerous). QRL 2.0/Zond moves to ML-DSA-87 + PoS, testnet only.	Single-asset coin plus EVM; the "PQ coin" niche is taken by them.
Nervos CKB	Opt-in SPHINCS+ lock script, audited, on mainnet since 2025. "Quantum Purse" wallet launched Feb 2026.	PQ is optional; most of the chain is still classical; SPHINCS+ signatures are large (8–50 KB).
Algorand	Falcon-signed state proofs; native Falcon accounts reported as phasing in (sources conflict).	Accounts are mostly still Ed25519.
Abelian	Lattice-based with ring-signature privacy (verify mainnet maturity).	Privacy-coin positioning; exchange-listing friction.

Takeaway: "post-quantum" on its own is not unique any more. QRL owns that headline. A combination of things nobody else offers together is still open.

Native multi-asset ledgers (issued coins without smart contracts)

Cardano: multi-asset UTXOs, with minting governed by policy scripts. Mature, but classical crypto.
Liquid (Elements): issued assets plus Confidential Transactions, but a federated sidechain with 15 functionaries.
Taproot Assets / RGB: assets live off-chain and are anchored to Bitcoin. Holders must keep proofs, so the UX is hard.
Tether is shipping USDT on RGB, which shows demand for asset issuance on UTXO rails.

None of these is post-quantum.

2. Where Arrekta can be unique

Today's core already has three properties no live network combines:

Post-quantum from genesis, mandatory, not opt-in. ML-DSA (FIPS 204) on every input. There are no legacy coins to migrate, so there is no "Q-day" exposure and no BIP-361-style confiscation debate.
Native multi-asset issuance with PQ authority. Stablecoins and RWA tokens get ticker uniqueness, supply caps and mint authority enforced by consensus, with no contract risk. This is Cardano/Liquid-style issuance, but quantum-safe.
Quantum-safe networking, not just quantum-safe signatures. P2P traffic uses hybrid X25519 + ML-KEM-768 encryption, so recorded traffic can't be decrypted later. Most chains send P2P traffic in plaintext; Bitcoin's BIP-324 encryption is classical only.

Proposed positioning

"The quantum-safe settlement layer for issued assets." A public PoW chain where stablecoins, tokenised deposits and RWAs can be issued and settled today with NIST-standard post-quantum cryptography, instead of waiting 5–10 years for Bitcoin or Ethereum to migrate.

This targets a real, compliance-driven buyer: issuers and institutions facing PQC mandates. Retail traders are a crowded market.

Differentiators to add (ranked by value ÷ effort)

#	Feature	Why it's unique	Effort
1	Crypto-agility built into consensus. A scheme registry plus a defined scheme-sunset procedure (announce → migration window → retire), specified from genesis.	Bitcoin is fighting over this now (BIP-361). We define it before launch.	Low (the scheme byte already exists)
2	Issuer controls for regulated assets. Optional per-asset freeze/clawback authority, declared at issuance and visible on chain so holders know the rules.	What stablecoin and RWA issuers need. Opt-in per asset; ARRK itself never has it.	Medium
3	Vaults and recovery keys. Time-delayed spends with a cancel key (covenant), giving theft protection for treasuries.	No PQ chain offers this. Bitcoin vault covenants are still debated.	Medium (Phase 7 script v2)
4	Signature-size engineering. Witness pruning for old blocks, then STARK-aggregated signatures per block (hash-based, so PQ).	Size is the #1 cost of PQ signatures. Solving it is a research moat.	High (research)
5	Atomic swaps with BTC/ETH (HTLCs). Move value in without a custodial bridge.	Avoids the bridge hacks that dominate crypto losses.	Medium
6	SLH-DSA as a second scheme. Hash-based fallback if lattice assumptions weaken.	Conservative dual-family security. CKB only has SPHINCS+, QRL only has XMSS/ML-DSA.	Low once pyca supports it; otherwise a vetted library
7	Confidential amounts.	Classical Pedersen commitments lose their binding property under quantum attack, which would make hidden inflation possible, so this must be lattice-based.	Research only; not v1

What to avoid: claiming "quantum-proof" (marketing) when we mean "uses NIST PQ standards"; privacy-coin positioning (exchange delistings); custodial bridges.

3. What we can extract (and legally)

Licence map

Project	Licence	Reuse verdict
Bitcoin Core	MIT (verify)	✅ Algorithms and code: addrman (eclipse resistance), mempool policy, compact blocks, headers-first sync, fee estimation
Elements / Liquid	MIT (verify)	✅ Asset issuance and reissuance design (ours already follows it)
rusty-kaspa	ISC (verify; one crate lists LGPL)	✅ for ISC parts: high-throughput Rust P2P, storage, GHOSTDAG/BlockDAG if we ever want faster blocks
reth	MIT/Apache-2.0 (verify)	✅ Staged sync, database layer (MDBX), node architecture
rust-libp2p, RocksDB	MIT / Apache-2.0	✅ Networking, storage
Polkadot SDK	Runtime/FRAME Apache-2.0, client GPL-3.0 + classpath	⚠️ Runtime pallets OK. The node client is GPL: our node would have to be GPL
Cosmos SDK / CometBFT	Apache-2.0	✅ Only if we switch to PoS/BFT, a different product. CometBFT validator keys are Ed25519, so they'd need PQ replacement
pyca/cryptography, OpenSSL	Apache/BSD	✅ Already used for ML-DSA, ML-KEM, ChaCha20-Poly1305
BCH ASERT, BIP-350 reference	MIT / BSD	✅ Already used

Rule: copy only MIT/Apache/ISC/BSD code, keep their copyright notices (THIRD_PARTY_NOTICES.md), and record where every file came from. GPL is fine only if we accept GPL for the node.

Three ways to go faster

Path	Speed to mainnet-grade	Uniqueness	Risk
A. Fork Elements/Bitcoin Core (C++) and add ML-DSA	Fast for P2P, mempool and storage maturity	Low: "another Bitcoin fork with PQ opcodes" (BTQ already did this on testnet)	Huge C++ codebase; ECDSA assumptions everywhere; team is Python-first
B. Build on Polkadot SDK or Cosmos SDK	Fast for PoS/governance	Low–medium: hundreds of appchains look the same	Changes the product to PoS; GPL client (Polkadot); classical validator keys
C. Keep our protocol, harden it with extracted parts ✅	Medium	High: our own spec, PQ + assets + PQ transport	We own the protocol, so we need the external audit (already in the plan)

Recommendation: C.

Keep core/PROTOCOL.md and the Python node as the reference implementation. It's done, tested, and every rule has a test.
Build the production node in Rust, ported from the spec. Lift permissively licensed parts:
- rust-libp2p, or our ML-KEM transport on tokio
- MDBX/RocksDB storage patterns from reth and kaspa
- addrman and compact-block algorithms from Bitcoin Core
- ML-DSA via aws-lc-rs/OpenSSL bindings
Cross-check both implementations against shared test vectors. Differential testing is the best defence against consensus bugs, and it is how serious chains run multiple clients.
Spend our own engineering time only where we're unique: assets, crypto-agility, issuer controls, vaults, signature-size research.

4. Decisions (taken 2026-10-18 by Moses)

Positioning: the quantum-safe settlement layer for issued assets (stablecoins, tokenised deposits, RWAs).
Consensus: stay proof-of-work (SHA3 + ASERT).
Issuer controls: opt-in per asset. Freeze/clawback authority is declared at issuance and visible on chain; ARRK never has it.
Production node: Rust port of the spec, extracting MIT/Apache/ISC components. The Python core stays as the reference implementation, and both are cross-tested with shared vectors.