Major consensus families
- Proof-of-work (PoW) — miners compete to solve a computational puzzle. The winner produces the next block. Security comes from the economic cost of electricity + hardware. Bitcoin, Dogecoin, Litecoin, Ethereum Classic.
- Proof-of-stake (PoS) — validators lock tokens as collateral. The protocol pseudo-randomly selects producers. Misbehavior is slashed. Ethereum (post-Merge), Solana, Cardano, Cosmos chains.
- Delegated proof-of-stake (DPoS) — token holders vote for a small set of active validators. Trades decentralization for throughput. EOS (historically), Binance Chain.
- Byzantine Fault Tolerant (BFT) — a fixed set of validators votes on each block, reaching finality quickly. Used in Cosmos (Tendermint), Algorand, Aptos, Sui.
- Proof-of-history (PoH) — Solana’s innovation. Each validator proves the passage of time via a sequential hash, letting the network order transactions before consensus. Combined with PoS for validator selection.
Design tradeoffs
Every consensus design balances three properties:
- Decentralization — how many independent parties can meaningfully participate.
- Security — how expensive is a successful attack.
- Throughput / finality — how many transactions per second, how fast is finality.
Bitcoin optimizes hard for decentralization + security; it’s slow (10-minute blocks, 1-hour effective finality). Solana optimizes for throughput (2,000+ TPS, sub-second finality) at the cost of more centralized validator requirements. Ethereum’s PoS targets the middle: ~15 TPS base layer, 12-15 minutes to economic finality, ~900,000 validators.
Attack economics
The security model comes down to: “how much would it cost to attack this chain?”
- Bitcoin — 51% attack requires renting/buying enough hashpower to outrun the network. Current cost: ~$20B to sustain for 24 hours.
- Ethereum — attacking requires controlling 33% (stall) or 66% (reorganize) of staked ETH. At $100B total stake, that’s tens of billions in capital + slashing risk.
- Smaller chains — attack costs on long-tail PoW chains can be as low as $1k/hour via hash-rate rental markets (NiceHash). Several small chains have been 51%-attacked historically.
Risks and considerations
No consensus is perfectly attack-proof. The practical security lives in the economic and social costs of attacks. For high-value activity (treasury holdings, institutional settlement), stick to chains where an attack is clearly uneconomic: Bitcoin for value storage, Ethereum for smart contracts, Solana for high-throughput apps.
For builders, the consensus choice is architectural — it shapes what applications can run, what finality guarantees are available, and how the chain degrades under stress. Most new chains are forks of existing consensus codebases rather than novel designs, because getting consensus right is notoriously hard and mistakes can be fatal.