One key, one machine, one failure
A conventional validator concentrates a stake, a key and an uptime obligation onto one operator. That operator is a single point of failure for its own stake and, in aggregate, a source of correlated failure for the network: when many validators share a host, a client or a data centre, they go down together.
The problem is not that operators are careless. It is that the unit of participation was drawn around a machine instead of around a duty.
The unit of participation was drawn around a machine instead of around a duty.
Splitting the duty, not the stake
DVT distributes a validator's signing duty across several nodes that must cooperate to act. No single participant holds the key, and no single participant failing takes the validator offline — the remaining members carry the duty.
This is the same instinct that runs through Hashi's cross-chain security work: prefer a design where the failure of one component is survivable to a design where every component must be trusted to behave.
