Problem: In the same way that regranting decentralizes grantmaking, so do the same thing for bounties. For example, give the top 20 AI safety researchers up to $100,000 to create bounties or RFPs for, say, technical research problems. They could also reallocate their budget to other trusted people, creating a system of decentralized trust.
In theory, FTX’s regrantors could already do this with their existing budgets, but this would encourage people to think creatively about using bounties or RFPs.
Bounties are great because you only pay out if it’s successful. If hypothetically each researcher created 5 bounties at $10,000 each that’d be 100 bounties—lots of experiments.
RFPs are great because it puts less risk on the applicants but also is a scalable, low-management way to turn money into impact.
Examples: 1) I’ll pay you $1,000 for every bounty idea that gets funded 2) Richard Ngo
Bounty Budgets
Like Regranting, but for Bounties
Problem: In the same way that regranting decentralizes grantmaking, so do the same thing for bounties. For example, give the top 20 AI safety researchers up to $100,000 to create bounties or RFPs for, say, technical research problems. They could also reallocate their budget to other trusted people, creating a system of decentralized trust.
In theory, FTX’s regrantors could already do this with their existing budgets, but this would encourage people to think creatively about using bounties or RFPs.
Bounties are great because you only pay out if it’s successful. If hypothetically each researcher created 5 bounties at $10,000 each that’d be 100 bounties—lots of experiments.
RFPs are great because it puts less risk on the applicants but also is a scalable, low-management way to turn money into impact.
Examples: 1) I’ll pay you $1,000 for every bounty idea that gets funded
2) Richard Ngo