BLK 01Foundations · 🟢 everyone · ≈6 h · 8 lessons + playground

Blockchain foundations —
and where it came from

By the end of this module you can explain a blockchain to a regulator without hand-waving: what a hash is, why blocks chain, how strangers agree, and why "immutable" comes with asterisks. We start with the 44-year story that got us here.

LO1 · explain blocks, hashing, consensusLO1 · sketch the history 1982→2026

L1A brief history: six eras

Blockchain was not invented in 2008. Bitcoin assembled four older ideas — digital cash, hash-chained timestamps, proof-of-work, and peer-to-peer networks — into one system that finally worked without a central operator.

1982–1997 · PrecursorsDavid Chaum proposes untraceable digital cash (eCash). Haber & Stornetta build tamper-evident hash-chained timestamps for documents (1991) — literally a chain of hashes. Adam Back's Hashcash (1997) prices email with proof-of-work.
2008–2012 · BitcoinSatoshi Nakamoto's whitepaper (Oct 2008) and the genesis block (Jan 2009) solve double-spending with no bank. First real purchase 2010. It runs, unhacked, to this day.
2013–2015 · ProgrammabilityVitalik Buterin's Ethereum whitepaper (2013): a blockchain that runs programs (smart contracts). Mainnet July 2015. The Linux Foundation launches Hyperledger; R3 designs Corda for regulated finance.
2016–2019 · Enterprise waveThe DAO hack tests Ethereum's immutability. ICO mania. Insurance and banking pile in: B3i, we.trade, Insurwave, AXA fizzy, State Farm–USAA subrogation pilots, MetLife Vitana.
2020–2023 · Boom & shakeoutDeFi and NFT waves; Ethereum switches to proof-of-stake (The Merge, 2022, cutting energy use ~99.95%). Terra and FTX collapse. The consortium era ends: B3i, we.trade, Marco Polo, Contour all shut down.
2024–2026 · Institutional eraUS spot ETFs; EU MiCA in force; US GENIUS Act (2025) regulates stablecoins. RWA tokenization hits $23 B (H1 2025). JPMorgan puts a deposit token on a public L2; Aon settles insurance premiums in stablecoins (2026).
Fig 1.1 Six eras, 1982→2026. Notice the rhyme: every boom funds infrastructure; every bust teaches a governance lesson; each cycle ends closer to regulated finance.
✦ The pattern that matters for insurers

The 2016–19 enterprise wave failed mostly on business models, not technology (Module 08/09 does the autopsies). The 2024–26 wave is different in kind: regulation exists (MiCA, GENIUS Act), the instruments are institutional (tokenized deposits, digital bonds), and the players are JPMorgan and Aon rather than startups. That's why this course exists now.

L2The problem: shared truth without a referee

Every financial system is a ledger — a list of who owns what, and what happened when. The traditional fix for disagreement is a trusted central party: a bank clears payments; an insurer's core system is the single source of truth for policies. That works, at three costs:

A blockchain is a replicated, append-only ledger where the participants themselves — not a referee — agree on each new entry, and where rewriting old entries is made computationally or politically expensive. Everything else in this module is the machinery that makes those two properties real.

L3Hashing — the fingerprint function

A cryptographic hash function (this course uses SHA-256) turns any input into a fixed 64-hex-character fingerprint with four properties:

PropertyMeaningWhy insurance cares
DeterministicSame input → same hash, alwaysAnyone can re-verify a record's fingerprint years later
AvalancheChange one character → a completely different hashSilent edits to a claim record are unhideable
One-wayCannot recover the input from the hashYou can publish a fingerprint of PII without publishing the PII
Collision-resistantInfeasible to find two inputs with the same hashA fingerprint uniquely commits to one document
◐ Galileo lens

Hash-as-commitment is the cheapest blockchain idea to adopt: hash a policy document or an AI decision log, anchor the hash, keep the data in your normal database. Galileo's audit ledger and Module 11's auditable-AI pattern are both elaborations of this one trick.

L4Blocks, chains, and Merkle trees

A block is a batch of transactions plus a header. The header contains the hash of the previous block — which is what turns independent batches into a chain: edit any historical block and every later block's hash breaks.

Block 41 header prev: 00f3…9cnonce · time · merkle root transactions premium P-1001 · 1,250claim C-77 · approved Block 42 header prev: hash(41) ✓nonce · time · merkle root transactions policy P-1002 issuedpayout C-77 · 50,000 Block 43 header prev: hash(42) ✓nonce · time · merkle root transactions party K-88 updatedpremium P-1002 · 980 hash links Tamper with a transaction in Block 41 → its hash changes → Block 42's "prev" no longer matches → Block 43 breaks too. History is welded together. The merkle root is a single hash summarising all transactions in the block — proof that a specific transaction is included, without shipping the whole block.
Fig 1.2 Block anatomy. The two load-bearing fields are prev (chains history) and the merkle root (summarises contents).

A Merkle tree pair-wise hashes transactions upward to a single root. Its gift is the Merkle proof: you can prove "claim C-77 is in block 42" with just ~log₂(n) hashes — the basis of light clients, and of audit-friendly designs where a regulator verifies inclusion without receiving your whole book.

L5Interactive: break a chain with your own hands

This is a live 4-block chain computing real SHA-256 in your browser. The rule: a block is valid when its hash starts with 00.

✦ What you just learned by feel

1. The avalanche effect — one character changed everything. 2. Chaining — corruption propagates forward, so old history is the hardest to fake. 3. Mining is just a brute-force search for a lucky nonce; "difficulty" = how many leading zeros you demand. On Bitcoin the target is astronomically harder and re-mining the whole chain would out-cost nation states — that's the security budget.

L6Keys, signatures, wallets

Ownership on a blockchain is cryptographic, not administrative:

△ The enterprise translation

For an insurer, "who holds the keys" becomes an operational-risk question: HSMs, key ceremonies, m-of-n multi-signature approvals, and recovery procedures. Custody is a bigger project than the chain itself — one reason permissioned platforms (Module 03) put identity management first.

L7Transactions & state: UTXO vs accounts

Bitcoin style

UTXO model

The ledger is a set of unspent coins ("unspent transaction outputs"). A transaction consumes some UTXOs and creates new ones — like paying with physical notes and getting change. Simple to verify in parallel; awkward for rich shared state.

Ethereum style

Account model

The ledger is a mapping of accounts → balances + contract storage, updated by transactions. Feels like a bank database, which is why business logic (smart contracts, Module 02) lives here. Corda, interestingly, went UTXO-like for privacy reasons.

L8Consensus — how strangers agree

Consensus answers one question: whose block is next? The answer differs by how much you trust the participants.

FamilyHow it picks the writerTrust settingCost profileExamples
Proof of WorkWhoever wins the hash lottery (mining)Fully open, adversarialEnergy-heavy; probabilistic finalityBitcoin
Proof of StakeValidators stake capital; misbehaviour is financially "slashed"Open, adversarial~99.95% less energy than PoW; fast finalityEthereum since 2022
BFT voting (PBFT/IBFT/Raft-like)Known validators vote; tolerate up to ⅓ faulty/maliciousIdentified members (consortium)Cheap, instant finality; needs governanceFabric ordering, Besu IBFT, Corda notaries
◐ Galileo lens

Galileo runs 3 blockchain nodes that it fully controls — so consensus there is not adversarial at all; it's closer to a replicated append-only log with BFT-style ordering. That's typical for the audit-ledger pattern: the chain's value is tamper-evidence and ordering, not decentralised trust. Ask "who are the validators, and who governs them?" of every enterprise chain you meet.

L9Public, consortium, private — and the immutability asterisk

Public

Anyone joins

Ethereum, Bitcoin. Maximum neutrality & auditability; least privacy and control. Insurance fit: parametric retail, crypto-native cover, anchoring proofs.

Consortium

Members join

Fabric, Corda, Besu networks run by an industry group. Shared workflows between competitors (claims, KYC, reinsurance). Fit: most enterprise insurance cases — if governance holds (see B3i).

Private

One operator

Single-company chain, like Galileo's. Tamper-evident audit within one trust domain. Honest question to ask: would a signed append-only database do?

△ "Immutable" — the asterisks

1. Immutability is economic/political, not physical: a public chain can fork (Ethereum did after the DAO hack); a 3-node private chain can be rewritten by its operator. 2. Immutability of garbage is garbage forever — data quality still matters. 3. GDPR/PDPA right-to-erasure means PII goes off-chain; the chain keeps hashes. Design for these from day one.

Watch

Sources for this module: Antonopoulos & Wood, Mastering Ethereum (concepts); Peng, Blockchain for Big Data (architecture layers, consensus taxonomy); public history records. Energy figure: Ethereum Foundation post-Merge estimate.