Crypto glossary
Nothing-at-stake problem
The nothing-at-stake problem is the risk that proof-of-stake validators sign every competing fork at no cost, since voting needs no real-world resources.
A proof-of-work miner who splits hash power across two forks earns less on each, so the cost of electricity pushes miners to back a single chain. In early proof-of-stake designs, signing a block on one fork did not prevent signing on another, and doing so cost nothing. The rational strategy was to vote on every fork to collect rewards whichever won, which would make it hard for the network to converge and easier to attempt double-spends.
The problem shaped how modern proof of stake works. Ethereum and most current designs require validators to lock stake that can be slashed, and they treat signing two conflicting blocks or attestations at the same height as a provable offence. Anyone can submit the two signatures as evidence, and part of the offender's deposit is destroyed. Finality gadgets add rules that make reverting finalized blocks extremely costly.
Slashing turns nothing at stake into something at stake, but only while the deposit is still locked; once stake is withdrawn, old keys again become free to misuse, which is the root of long-range attacks. Slashing can also hit honest operators who run the same keys on two machines by mistake.
Example
A validator runs a backup node with the same keys as its main node; both sign different blocks for the same slot, and the network slashes part of its staked ETH as if it were attacking.