weak subjectivity

Weak subjectivity is the honest admission that a proof-of-stake node cannot safely bootstrap from nothing but the protocol rules and the genesis block — it needs one piece of recent, externally-trusted information to start from. Proof of work is objective: a brand-new node can download every chain it sees and pick the one with the most cumulative work, with total confidence and no outside help. Proof of stake cannot, because the cheapness of signing makes a forged old chain look just as valid, so a small, bounded act of trust is required at sync time.

That trusted input is a weak subjectivity checkpoint: a recent finalized block root and slot, distributed out-of-band — baked into the client software, fetched from a block explorer, or received from a trusted peer. A syncing node treats this checkpoint as ground truth and refuses any chain that does not contain it, which neutralizes long-range attacks regardless of how internally consistent the fake chain is. The trust is 'weak' because it is minimal, falsifiable, and only needed once: after syncing past the checkpoint, the node verifies everything else cryptographically on its own.

There is a weak subjectivity period — the span over which a checkpoint stays safe to use — derived from how fast the validator set can churn: roughly, how long it would take for enough stake to exit that an attacker could assemble a supermajority of now-stake-free keys. For Ethereum mainnet with very large total stake, that period is on the order of weeks to months, so a checkpoint only a few weeks old is more than sufficient. The practical guidance is simple: when spinning up a fresh node, fetch a recent finalized checkpoint from a source you trust rather than syncing blindly from genesis.