The ability of a network to keep a single honest history even if some nodes lie or go offline.

A replicated data structure designed so concurrent changes can converge under specified merge or delivery rules, without coordinating every update.

The problem of getting separate processes to agree on values despite failures, subject to explicit communication and fault assumptions.

A consistency guarantee under which replicas converge once updates stop and propagate, while intermediate reads may disagree.

The rule a node applies to select its preferred chain head when more than one valid branch is available.

An Ethereum proof-of-stake recovery mechanism that increases penalties for inactive validators when the chain fails to finalize for an extended period.

The longest or heaviest-chain rule introduced by Bitcoin. Probabilistic finality, open miner set.

Consensus run by a known, permissioned set of validators. Fast, centralized, common on sidechains.

Consensus where validators lock tokens as collateral to propose and attest blocks. Used by Ethereum.

Consensus where miners expend computation to propose blocks. Used by Bitcoin.

A method of replicating a service by having deterministic replicas process the same ordered inputs and produce corresponding states and outputs.

The need for a proof-of-stake node joining or returning after a long absence to obtain a sufficiently recent trusted checkpoint before validating forward.