Proof of Authority

Proof of Authority (PoA) is a consensus mechanism used mainly in permissioned blockchains, where a small set of pre-approved, publicly identified validators take turns producing blocks. Rather than staking energy or capital, these authorities stake their real-world identity and reputation, which is forfeited if they misbehave. The term was popularized around 2017 by Gavin Wood in the Ethereum tooling world. It is included here for completeness and contrast: it is a deliberately centralized design, unsuited to Bitcoin's permissionless goals, but understanding why sharpens the case for proof-of-work.

How validators are chosen

Validator status is granted through vetting: candidates typically must disclose a verifiable identity, demonstrate good standing, and be admitted by whatever governance the network runs. Because the validator set is small and known, blocks are produced quickly and cheaply, often in round-robin order, with finality reached in seconds and without resource-intensive competition. Throughput is excellent and energy use negligible — when you already trust the operators, expensive consensus is redundant. This makes PoA a rational fit for consortium networks, supply-chain ledgers, and test networks; Ethereum's long-running testnets used PoA precisely because test ether is worthless and attack incentives are absent.

The decentralization trade-off

PoA's efficiency comes from concentrating trust in a handful of named parties, and that concentration is its core weakness for public money. A few validators can censor transactions or collude; the network's integrity depends on the honesty and continued availability of specific operators; and — decisive for anything with monetary value — identified authorities are addressable. A regulator, court, or armed authority can find them, compel them, or shut them down, which means the system's censorship resistance is exactly that of its most pressurable validator. Note too that "identity staked" is a soft bond: a reputation, unlike burned energy or slashed collateral, cannot be objectively priced or automatically confiscated by the protocol. Requiring identity also forfeits Sybil resistance through cost, replacing it with Sybil resistance through gatekeeping — the very permission Bitcoin exists to remove.

The taxonomy is worth keeping straight, because "PoA" gets applied loosely. Delegated and nominated staking systems where a small elected validator set produces blocks are economically different but operationally similar — a couple dozen known parties in a room — and inherit much of the same censorship surface. Even within permissioned deployments, resilience varies with validator diversity: five validators inside one company is a database with extra steps, while thirty across independent institutions and jurisdictions at least distributes the pressure points. The analytical habit transfers to evaluating any chain: count the block producers, ask who can identify them, and ask what it costs — in energy, capital, or reputation — to become or to coerce one.

The Bitcoin contrast

Bitcoin's answer to the same problem inverts every choice PoA makes. Anyone may mine, anonymously, from anywhere, and influence is purchased only with ongoing, externally verifiable expenditure of energy — no admission committee, no identities, no one to subpoena. That is what Nakamoto Consensus buys with its energy budget: validators who cannot be enumerated cannot be co-opted as a group. The steelman for PoA is that it never claims otherwise — it is an efficiency design for contexts where the trust already exists, essentially a replicated database with signatures, and judged as that it is perfectly sound engineering. The error is only in offering it as a peer of permissionless systems; every open consensus design must earn Byzantine Fault Tolerance against unknown adversaries, which is a different and harder problem than coordinating known colleagues.

For permissionless alternatives and their trade-offs, see Proof of Stake. The one-line summary worth keeping: PoA answers "how do trusted parties agree efficiently?" — Bitcoin answers "how do strangers agree without trusting anyone?" Only the second question needed inventing.