Blockchain — Deep Dive
Internal Data Structures
Merkle Trees
Each block doesn’t just list transactions — it organizes them in a Merkle tree (binary hash tree). This allows:
- Efficient verification: Prove a transaction is in a block by providing ~log₂(n) hashes instead of all transactions
- Light clients: Mobile wallets don’t download full blocks — they verify Merkle proofs
Root Hash
/ \
Hash(AB) Hash(CD)
/ \ / \
H(A) H(B) H(C) H(D)
| | | |
Tx A Tx B Tx C Tx D
UTXO vs Account Model
- Bitcoin (UTXO): Unspent Transaction Outputs. Each “coin” is a discrete object. Privacy-friendly but complex.
- Ethereum (Account): Each address has a balance. Simpler but requires careful nonce management.
Consensus Deep Dive
Byzantine Fault Tolerance (BFT)
The fundamental problem: in a network of n nodes, up to f can be malicious (lie, delay, send conflicting messages). BFT consensus works if f < n/3.
Nakamoto consensus (Bitcoin) is a probabilistic BFT — finality is never absolute but becomes exponentially certain with each confirmation (6 blocks ≈ 99.9999% final).
Proof of Stake Economics
Ethereum’s PoS requires validators to stake 32 ETH. Misbehavior triggers slashing — loss of staked funds. This creates economic incentive alignment:
- Honest behavior: Earn ~4-5% APY
- Dishonest behavior: Lose staked ETH (up to 100%)
The Scalability Trilemma
You can optimize for two of three:
Decentralization
/ \
/ \
Security --- Scalability
- Bitcoin: Decentralized + Secure, but ~7 tx/sec
- Solana: Scalable + Secure, but fewer validators (less decentralized)
- Layer 2s (Rollups): Inherit L1 security while scaling to thousands of tx/sec
Rollups
- Optimistic Rollups: Assume transactions are valid, allow fraud proofs within a challenge period (7 days). Used by Arbitrum, Optimism.
- ZK-Rollups: Cryptographic validity proofs included with each batch. Faster finality but more complex. Used by zkSync, StarkNet.
Gas and Fee Markets
Ethereum’s EIP-1559 introduced:
- Base fee: Algorithmically adjusted per block. Burned (deflationary pressure).
- Priority fee (tip): Paid to validators for inclusion priority.
- Fee = (base_fee + tip) × gas_used
This creates a more predictable fee market compared to first-price auctions.
Smart Contracts
Self-executing code deployed on-chain. Key considerations:
- Immutable once deployed — bugs can’t be patched (unless using proxy patterns)
- Gas costs real money — every computation has a price
- Composability — contracts can call other contracts (DeFi “money legos”)
- Reentrancy attacks — the most common vulnerability (The DAO hack, 2016)