Where he's right:

The incumbency-by-inertia diagnosis is real. Stake-weighted approval voting with a "return proposal" floor genuinely does reward renewal over scrutiny, and autopilot whale votes are a well-known failure mode. The milestone-vs-drip point is his strongest structural argument - paying daily regardless of shipping is a weak incentive design, and escrow/bounty models do align better for discrete deliverables. Public numeric reporting and the newcomer cold-start problem (scepticism lands hardest on people who haven't yet had a chance to burn anyone) are both fair.

Where the thinking has holes:

The biggest is causal. "We spent millions and the ecosystem is smaller, therefore the spending didn't work" ignores the counterfactual. Nobody knows the state of Hive without that spending; it could plausibly be worse. And "plenty of chains grew during the same period" is survivorship bias - the chains that grew mostly did it on VC injection, token emissions, and speculation cycles Hive structurally doesn't have. He picks the comparison that flatters the argument.

Second, he conflates categories he finds boring with waste. "Another frontend" is dismissed, but frontends are the actual surface users touch, and redundancy there is a censorship-resistance feature, not fat. More pointedly, the "boring stuff that matters" he wants funded (onboarding that hides private keys, tooling, current docs) is largely built by exactly those frontend and infra teams he's criticizing. His preferred spending and his disfavored recipients overlap heavily.

Third, there's an internal tension between his novelty bias and his own solution. He wants new, unproven people funded because they "have something to prove," but the escrow tool he built exists precisely because new people can't be trusted on reputation. Unknown builders taking small grants and vanishing is the classic grift vector, and small proposals are harder to police than large ones, not easier. His "8 of 10 fail is fine" model has no accountability mechanism for the 8.

Fourth, the anti-renewal stance contradicts the infrastructure argument. Load-bearing infrastructure (RPC nodes, core dev, security, the docs and tooling he wants) is exactly the kind of thing that needs stable, perpetual, renewing funding. A hard two-year cap would threaten the maintainers of the unglamorous stuff he claims to prize.

Fifth, and most practically: his reforms have no enforcement mechanism. The DHF is just stake-weighted approval voting. There's no allocator who can "reserve a portion for newcomers" or "cap renewals." Every fix he proposes would require the same whales whose autopilot voting he's criticizing to voluntarily constrain their own discretion, or a consensus protocol change. He diagnoses the mechanism correctly and then proposes solutions the mechanism can't implement.

Net: he's directionally right about the governance pathology and wrong, or at least overconfident, about the conclusion that spending "failed" and that novelty and newcomers are the fix. It reads more as a well-argued provocation than a costed reform plan.