The earlier PoA/PoS comparison mixed validator selection, consensus mechanics, governance, and performance into one ranking. Separating those dimensions makes the trade-offs easier to inspect.
| Question | Permissioned authority model | Stake-based model |
|---|---|---|
| Who can validate? | An authorized validator set | Rules admit or select validators using stake |
| What controls influence? | Membership and voting rules | Stake weight and protocol rules |
| What discourages abuse? | Protocol checks plus the system’s authority/governance model | Protocol-specific penalties and economic exposure |
| What determines finality? | The actual consensus protocol and assumptions | The actual consensus protocol and assumptions |
Permissioned does not specify the whole protocol
Besu documents IBFT 2.0 and QBFT as PoA options with explicit validator and quorum requirements. A PoA label alone does not establish Byzantine fault tolerance, immediate finality, governance concentration, or a throughput number. Reference: Besu PoA concepts.
Stake changes the participation and incentive rules
Ethereum’s PoS uses staked ETH, proposer/attester duties, and penalties. Its finality rules are specific to that protocol; they should not be assumed for every system called PoS. Reference: Ethereum PoS.
Open admission does not guarantee dispersed control: stake, infrastructure, delegation, and governance can concentrate. Likewise, a small permissioned set does not remove network partitions or operational failures.
Both categories can avoid proof-of-work mining, but neither label supplies an automatic scalability or decentralization ranking. The old example-chain catalog and setup-tool recommendations are omitted rather than presented as current guidance. No consensus network was created or benchmarked.