Delegated Proof of Stake

Why It Matters

If every validator has to process every block, the network is only as fast as its slowest participant. DPoS makes a blunt trade: elect a small, known roster of block producers, and the chain gets fast. The question it forces — and the reason it’s worth studying even if you never use a DPoS chain — is whether a system with twenty-one elected operators is meaningfully decentralized at all.

The casebook’s critique (via Micali) is unusually sharp: DPoS is centralization “by declaration” — and because the same known parties hold power for a stretch, the delegate set is a standing denial-of-service target. Knowing exactly whom to attack is a gift to adversaries.

How It Works

Beginner

Think of it as a representative democracy for block production. Instead of every token holder validating, you vote — with voting power proportional to your tokens — for a small slate of delegates. The winners take turns producing blocks. If a delegate misbehaves or goes offline, voters can replace them at the next round.

Intermediate

Token holders continuously vote, stake-weighted, for block producers (typically ~21–101 depending on the chain). Elected delegates produce blocks round-robin on a fixed schedule, which removes the lottery entirely — block times are deterministic and fast. Rewards flow to delegates, who often share them with their voters, creating campaign-style competition based on reputation.

Two structural critiques follow directly: the validator set is small and publicly known (DoS target, collusion risk), and stake-weighted voting tends toward cartel equilibrium — large holders and exchanges can entrench a stable slate.

Builder

EOS: 21 block producers, 0.5s blocks, producers re-elected continuously by stake-weighted approval voting. Tron: 27 “Super Representatives,” 3s blocks. Both deliver thousands of TPS precisely because consensus is a scheduled rotation among few, well-provisioned nodes — the throughput is bought with the validator count. Note the casebook’s claim that no DPoS chain had launched is dated: EOS (2018) and Tron (2018) have run for years.

Examples

EOS — The canonical 21-producer DPoS chain.
Tron — 27 elected Super Representatives.
Lisk, BitShares, Steem — Earlier DPoS implementations from the same design lineage.

Tradeoffs

Strengths

High throughput, fast finality — few, known producers on a fixed schedule means no fork lottery.
Accountability — delegates are identifiable and removable by vote.
Energy-light — inherits PoS’s efficiency.

Limitations

Centralization by declaration — the casebook’s core critique: a small elected set is structurally centralized regardless of the election.
DoS target — known producers can be attacked precisely because everyone knows who holds power this round.
Voter apathy and cartels — low participation plus exchange-held stake tends to freeze the delegate slate.

Main alternative: Pure Proof of Stake — keeps PoS’s efficiency but replaces a standing elected set with secret, per-step random committees, eliminating the DoS target.

Proof of Stake — The parent mechanism
Pure Proof of Stake — The casebook’s explicit counter-design
Distributed Consensus — The problem all variants solve
51% Attack — Majority control in vote-based systems

Sources & Last Updated

MIT BLC Module 2: Maintaining Blockchain Integrity (primary source)
Vault note: Delegated Proof of Stake (M2 cluster)

Freshness note: the casebook’s “no DPoS chain has launched” claim was already outdated at publication of this page — corrected inline per the vault’s freshness flag.

Last updated: June 10, 2026