A bounty network has exactly one job: pay for real bugs, and never pay for fake ones. So we verify every finding twice — once by the verifier who checks it, once by the quorum that agrees.

The verifier re-runs the exploit in a sandbox and commits a hash of the exact input and output. The quorum votes on that hash, weighted by stake. Two independent checks, because a single check is a single point of failure — and a single point of corruption.

It costs a little latency. It's why the payout clears in minutes instead of milliseconds. Worth it. A bounty network that pays fake bugs is just a faucet for liars.