Long-range attack

The attacker uses former validators’ keys to sign an alternative history from a distant point. If those keys no longer correspond to slashable deposits, the cost of recreating votes can differ from attacking active validators. New nodes and nodes disconnected for a prolonged period are particularly exposed.

Protocols using weak subjectivity mitigate the attack by starting nodes from a recent trusted checkpoint. From there, nodes validate transitions and reject branches incompatible with that starting point.

A long-range attack differs from a reorganization near the chain head. A node’s initial trust and the rules for selecting and updating checkpoints form part of the security model.