Validator Slashing and Penalty Design

In a proof-of-stake blockchain, validators lock up ("stake") coins as a bond and are chosen to propose or confirm new blocks. If a validator breaks the protocol's rules, signing two conflicting blocks at the same height, or going offline for extended periods, the network can slash the bond, destroying some or all of the staked coins as punishment. The size and trigger conditions of that penalty are a deliberate design choice: too lenient, and dishonest behavior becomes cheap; too harsh, and honest validators get punished for accidental downtime, discouraging anyone from staking at all.

Worked example

A validator stakes 32 ETH. For a minor offense like brief downtime, the protocol might slash only a small fraction, roughly 0.001 ETH, enough to sting without ruining an honest operator who had a server outage. For deliberately signing two conflicting blocks (equivocation), the same protocol might slash 1 ETH or more immediately, plus force an early, penalized exit from staking, a penalty over a thousand times larger, reflecting that this behavior can only happen through deliberate malice or serious negligence, not bad luck.

Some designs add "correlation penalties" on top: if many validators equivocate at once, suggesting a coordinated attack rather than isolated error, the slash for each one scales up sharply, since simultaneous misbehavior across the network is a far stronger signal of genuine attack than any single validator failing alone.