That's a really fair point, and I think you've actually identified the exact same problem I'm describing, just from the other end.
You're right that project-based funding has burned Hive before. People pitching big ideas, collecting six figures, and delivering nothing or next to nothing. I've watched that happen too and it's frustrating. So I get why the instinct is to go smaller and more controlled, and honestly, I respect that this proposal is designed that way. Conservative budget, pilot scope, vetting applicants. That's all sensible.
But I think the lesson from those failed projects isn't that we should stop funding projects. It's that we were funding them wrong. The reason people walked away with $100k and no results is exactly what you said: no one was actively managing it from an investor perspective. No milestones, no stage gates, no "deliver phase one before you get phase two money." The accountability structure wasn't there, so the money just flowed until someone eventually noticed nothing was happening.
That's a fixable problem. If you fund a project in stages tied to deliverables, the worst case scenario is you lose one milestone payment before you cut someone off. That's a completely different risk profile than handing someone a year of funding and hoping for the best.
The other thing milestone and task-based funding does is open up a single feature or problem to multiple contributors. Right now, most proposals are one dev, one project. If that person gets busy with their day job, burns out, or just disappears, the whole thing stalls. But if you break a large feature into discrete tasks, three or four developers can work on different parts of the same problem in parallel. You're not betting everything on one person's availability and motivation. The work keeps moving even if one contributor drops off. That's just a more resilient way to build software, and it's how most serious open source projects operate outside of Hive already.
And what you described wanting for the future of Hive Dev Fund, a smaller scale system where tasks and small projects can get funding without the massive overhead of the DHF, that's basically what I'm arguing for too. I think we're on the same page about where this needs to go. I just think we should start there now instead of routing community funds through AI subscriptions first and evolving toward it later. The task-based model is the stronger foundation, and I'd rather see the pilot built around that from day one.