Just came across something Vitalik shared recently about Deep Funding that's actually pretty interesting. He's basically talking about combining two different approaches - and it's worth understanding what he means here.

So the first part is about value graphs. Instead of asking the impossible question "how much has X contributed to humanity", you flip it: "how much of Y's contribution actually comes from X?" It sounds like a small shift but it's actually clever. Measuring abstract contribution is notoriously hard - there's this whole problem where people will say they'd pay $80 to save 2000 birds but also $80 to save 200000 birds. Makes no sense. But asking "which matters more to C, A or B?" - that's something people can actually answer.

The second part is about refining human judgment through a deep community layer. Basically anyone can throw in suggestions using whatever method they want - AI models, algorithms, whatever. Then you have human juries that spot-check random samples of these suggestions. Whichever submission aligns best with what the jury thinks is right becomes the output.

What's interesting here is how it tries to solve the scaling problem. You can't have humans evaluate everything, but you also can't just trust pure algorithmic output. This middle ground - combining AI suggestions with targeted human verification - is something we're probably going to see more of in funding mechanisms going forward.

Vitalik's been thinking about these mechanisms for a while, and this feels like a natural evolution of those ideas. Worth paying attention to if you're interested in how communities might allocate resources more effectively in the future.