Skip to content

Other Consensus Mechanisms

Beyond proof-of-work and proof-of-stake, a range of consensus mechanisms trade decentralization for performance, or replace capital with other resources. This page covers the most prominent variants.

Delegated Proof-of-Stake (DPoS)

DPoS was popularized by Dan Larimer (BitShares, Steem, EOS). Token holders vote for a small set of elected block producers rather than directly participating in consensus.

EOS structure:
21 active Block Producers (BPs)
Elected by token holder votes (1 token = 1 vote)
BPs take turns producing blocks in a rotating schedule
Block time: 0.5 seconds
Throughput: ~4,000 TPS

The small, known validator set enables high throughput and fast finality. The cost is decentralization: 21 BPs is a political structure, not a trustless one.

Observed failure modes in practice:

  • BP cartels: EOS block producers coordinated to vote for each other’s nodes (vote trading), undermining competitive election.
  • Voter apathy: most token holders do not vote. Whale accounts dominate elections.
  • Governance capture: the EOS “freeze” in 2018, where the EOS Core Arbitration Forum ordered transactions reversed, demonstrated that 21 elected parties can collectively make politically charged decisions.

TRON uses a similar 27-validator DPoS model.

Proof-of-Authority (PoA)

PoA replaces anonymous computational work or anonymous bonded capital with known, approved validators. Only whitelisted addresses can produce blocks.

Clique (Ethereum PoA, used in testnets):
A list of signers is maintained on-chain
Signers take turns proposing blocks in round-robin
A block is valid if signed by an authorized signer
Signers can vote to add or remove other signers

Where it is used:

  • Ethereum testnets: Goerli used Clique PoA; Sepolia uses a variation. Easy to reset and manage for developer testing.
  • Enterprise chains: Hyperledger Besu in consortium deployments, Quorum (JPMorgan’s Ethereum fork), Baseline Protocol implementations.
  • Polygon PoS uses a variant where validators are permissioned via a staking mechanism on Ethereum, but the active set is small enough to be quasi-PoA.

PoA is highly performant (block time can be sub-second, throughput thousands of TPS) but entirely dependent on the validator whitelist. A compromise of the validator keys or regulatory pressure on identified validator operators breaks safety.

Proof-of-History (PoH)

Proof-of-History is Solana’s solution to the “what time is it?” problem in distributed consensus. It is not a consensus mechanism on its own but a cryptographic clock that enables Solana’s Tower BFT to operate efficiently.

PoH sequence (continuous VDF):
h0 = SHA256("genesis")
h1 = SHA256(h0)
h2 = SHA256(h1)
...
hN = SHA256(h(N-1))
Between hashes, the leader inserts events (transactions):
hK = SHA256(h(K-1) || tx_data)
This proves that tx_data was created after h(K-1) and before h(K+1).

The continuous hash chain is a Verifiable Delay Function: it cannot be computed faster than sequentially, and each output proves that a specific wall-clock duration has elapsed since the previous output (since each SHA-256 takes finite time).

Why it matters for throughput: validators can independently verify the ordering of events without network round-trips to agree on time. This removes a bottleneck that typically limits BFT protocols to lower throughput.

Solana targets 400ms slots. The PoH leader streams a continuous sequence; validators verify in parallel.

Proof-of-Space (PoSpace) and Proof-of-Capacity

Chia Network (launched 2021) uses Proof-of-Space-and-Time. Instead of spending electricity to compute hashes, farmers pre-compute large lookup tables (plots) stored on disk. To win a block, a farmer responds to a challenge by looking up values in their plots.

Plotting (offline, one-time per disk):
Fill disk with cryptographic lookup tables
~100 GB per plot on an 8 TB drive = ~80 plots
Farming (online, continuous):
Network broadcasts a challenge
Farmer finds best matching value in plots
Submits proof; winner selected by best match
Expected win time proportional to total plot space

Proof-of-Time: to prevent grinding attacks (pre-computing many candidate plots and selecting the best after seeing the challenge), a Verifiable Delay Function runs after each plot selection to ensure plots cannot be computed on-the-fly.

Energy use is dramatically lower than PoW: a farmer can participate with a desktop PC and USB drives. The hardware cost is dominated by storage, not electricity.

Trade-offs: Chia’s launch caused a hard drive shortage as speculators purchased drives for plotting. The “grinding” attack surface (pre-computing many plots in advance) is an ongoing research concern.

HotStuff and its derivatives

HotStuff (2018, used in Diem/Libra) is a leader-based BFT protocol that achieves linear message complexity (O(n) messages per consensus round instead of the O(n^2) typical of PBFT).

Three-phase commit (simplified):
Leader broadcasts Prepare message
Replicas respond with votes
Leader collects 2/3 quorum, broadcasts Pre-commit
Replicas respond; leader broadcasts Commit
Replicas commit the block

The key insight: a threshold signature (BLS aggregation) lets the leader collect 2/3 votes and compress them into one message. This reduces network overhead from O(n^2) to O(n) for each consensus round.

HotStuff or variants appear in Diem (Facebook’s canceled stablecoin project), Aptos, and Sui.

Comparison of all mechanisms

MechanismChainFinalityThroughputPermissionlessEnergy
PoW (SHA-256)BitcoinProbabilistic~7 TPSYesVery high
PoW (RandomX)MoneroProbabilistic~1.7 TPSYesHigh (CPU)
PoS (Casper)Ethereum~12 min~15 TPSYes (32 ETH)Low
PoS (Tower BFT)Solana~400ms~50,000 TPSYes (hardware)Low
PoS (Tendermint)Cosmos1 block (~7s)~10,000 TPSYes (bonded)Low
DPoSEOS, TRON1 block (~0.5s)~4,000 TPSNo (elected)Very low
PoA (Clique)Test nets1 blockHighNo (whitelist)Minimal
PoSpaceChiaProbabilistic~52 TPSYes (storage)Low

Finality types

Probabilistic: the more blocks built on top of a block, the harder it is to revert. Never truly final, but practically final after sufficient depth (6 blocks for Bitcoin, ~35 for Ethereum pre-finalization).

Economic: a finalized block would cost > $1B to revert due to slashing. Ethereum post-Casper finalization.

Deterministic: a committed block is guaranteed to never be reverted given the security assumption holds. BFT-based systems (Tendermint, HotStuff). Requires a fixed known validator set.

References