Open questions

Mougayar's "Peak EF" thesis frames the Foundation as a transitional benefactor: it bootstrapped the commons because no one else could, and over time its share of network funding will fall. The question is what comes next. Validators are the obvious candidate substrate — they are the only set of actors with both a recurring revenue stream tied to the network and a credible claim to neutrality. But "validators fund public goods" is not a governance structure; it is a description of capital flows.

The first serious attempt is VRR — Validator Revenue Redistribution — proposed by Devansh Mehta and Clément Lesaege. It is elective and non-coercive: validators opt into routing a slice of rewards to a chosen pool. That removes the obvious objection (forced taxation by protocol) but inherits a different one: pitching for recurring funding requires a formalized decision body, and that body does not yet exist. Octant resembles a prototype, but it is one foundation's program, not a network-wide allocation mechanism.

Some kind of validator quorum, or staking-weighted vote, could be the substrate. The obvious risk is that it recreates plutocracy: whoever controls the most stake controls the most allocation. That is not "credibly neutral" — it is the EF problem with worse legitimacy. Reasonable answers probably look like quadratic schemes, randomized juries, or domain-specialized committees. We do not know which and neither does anyone else.

A validator that gave 500 ETH once last year ranks higher under absolute count than a validator giving 30 ETH every month for the same period. The second pattern is arguably more committed: it survived a year of governance, market conditions, and internal pressure to redirect that money elsewhere. The first might just be a press release. Our current methodology rewards the press release.

The fix is some kind of consistency score — perhaps a Gini-style measure of how evenly contributions are distributed across the window, or an exponential decay that weights recent and consistent giving over historical lump sums. Both have failure modes. Decay penalizes entities whose grant cycle is naturally annual. A consistency score penalizes entities responding to one-off opportunities like a critical security bounty.

For now we report absolute and percentage and let readers do the second derivative themselves. If you have a clean way to surface durability, write to us.

The three models are visible in the current dataset. Lido's 1% to Protocol Guild was a DAO vote: every staker who delegates accepts it implicitly, and dissent means withdrawing. That is closest to opt-out. Octant runs on Golem Foundation's own stake and is fully opt-in by the funder, with users choosing recipients. Pure protocol-default — a fraction of issuance hardcoded to a public goods address — has been proposed in various forms but is politically untouchable because it changes the social contract for stakers who never agreed to it.

Each path has a different governance and disclosure problem. Opt-out via DAO vote gives the appearance of consent without the substance: most delegators do not vote and many do not know about the redistribution at all. Opt-in is honest but produces small, lumpy flows. Protocol-default is the cleanest but requires a credibility for the recipient that no entity currently has.

Our intuition is that the answer is layered — a small protocol-default floor, a larger opt-out via staking pool DAOs, an opt-in tier on top — but the layering is a research project of its own.

Five entities — Lido, Coinbase, Kraken, Binance, Figment — operate or control a meaningful share of Ethereum stake. If those same five end up controlling most validator-allocated public goods funding, the network has not decentralized. It has swapped one funder for an oligarchy of funders, with worse alignment to the commons because four of them are profit-maximizing custodians and one is a stake aggregator with mixed incentives.

The walkaway test — can the network survive any one of them leaving? — passes today and probably continues to pass. But "the network can survive without you" is a low bar. The more interesting question is what governs the funding decisions made by whatever oligarchy emerges, and whether the legitimacy of those decisions survives contact with stakers who never consented to the funding choices being made on their behalf.

The strongest version of this argument is that any aggregation of stake into discretionary funding power is a step backward, and the right answer is protocol-level redistribution with no human in the loop. The weaker version is that aggregation is fine if the governance of the aggregated funds is itself decentralized. We find the strong version more philosophically clean and the weak version more pragmatic. We are not sure which is correct.