Protocol Guild is a collective funding mechanism for the people who maintain Ethereum's base layer. It describes itself as three components: an eligibility framework, a member registry, and a set of onchain donation contracts, designed so that "Ethereum's layer 1 R&D to be funded in the same way it is produced: a commons of peers and their collaborative effort over time."

It earns a place in this directory by being the structural opposite of almost everything else in it. Protocol Guild has no token. There is no supply to distribute, no market to price it, no holders to court, and no token-weighted vote to win. It has ~184 beneficiaries and $183m of lifetime inflows, and the beneficiaries cannot vote on the money. That last point is not an oversight. It is the design.

The money is deliberately ungovernable

Every DAO treasury in this wiki shares a property so basic it usually goes unnamed: a vote can move the funds. That is what a treasury is — a pot, and a process for deciding where it goes. Nearly every failure catalogued in how DAOs fail is downstream of it. If a vote can move the pot, then capturing the vote captures the pot, and everyone who might want the pot has a reason to try.

Protocol Guild removes the pot. Donations go to an immutable vesting contract that releases funds linearly, block by block, over four years. The docs are unusually blunt about what that word means (onchain architecture):

Here, "irrevocably" means donations cannot be stopped or otherwise redirected during the vest by anyone, be it the donor or Protocol Guild membership.

The membership page states the limit again from the members' side (membership):

Governance does not include any control over any vesting funds. e.g. members can't vote to cancel or change the timeline of vesting funds

So the members govern the mechanism — who is eligible, which contracts to use, how long vests should be, where to fundraise, what cut the legal entity takes — but they cannot govern the money already in flight. A donor cannot claw it back. A member majority cannot redirect it. There is no proposal that reaches it. What the members can do is decide who is on the list, and a formula does the rest.

Time weight: a formula instead of a vote

Each member's share of the Split contract is computed, not negotiated (time weight):

time_weight = SQRT((start_date − months_inactive) × full_or_part_time)
split_share = (time_weight / total_time_weights) × 100

The inputs are all facts about contribution, and none of them are money:

• start_date — when the member began contributing to Ethereum L1 R&D. Earlier means a larger share; the mechanism pays retroactively for history it did not fund at the time.
• months_inactive — tracked breaks. Existing members may take up to 3 months off without triggering a membership change.
• full_or_part_time — 1.0x for full-time (at least 40 hr/wk), 0.5x for partial (20–40 hr/wk). There is no third tier.
• SQRT — compresses the range, "to not overly privilege long-term members over newer contributors." The same square-root instinct that drives quadratic funding, applied to tenure rather than to donor counts.

Weights are republished onchain each quarter. The effect the docs highlight is that the formula dilutes its own incumbents: "Existing contributor weights get 'diluted' as newcomers show up." Adding a member costs every current member a slice of every future distribution — which is exactly the incentive the Guild wants its curators to feel, because the curators are the members.

Self-curation, and why the curators are the beneficiaries

Protocol Guild calls this self-curation: the membership selects its own beneficiary set. There is no grants council, no badgeholder cohort, no external reviewers — the structure that retroactive public goods funding depends on and that Optimism paused in January 2026. The docs defend the choice on four grounds, and the fourth is the sharpest:

It is incentive compatible that curators (Guild members) are drawn from the beneficiaries (Guild members). Adding beneficiaries removes future vested value from existing members. They will more carefully consider potential members and their contributions. An external council would not feel this constraint so directly.

They also argue that an external council is a capture surface of its own — "In the worst case, their goal is to maintain their position as curators - not to accurately curate the membership" — and that a curating body would need its own nomination and removal machinery, "more time, overhead, and bandwidth taken from the actual work of core protocol stewardship."

The counterweight against a self-dealing cartel is stated as an obligation rather than a hope: "The mechanism must accept all legitimate contributors... This prevents the set from ossifying or getting captured." The enforcement is social and procedural, not cryptographic. Additions and removals are pull requests against a public repo, open at least a week, merged on rough consensus. Removals are supposed to be self-removals. Peer removal exists for members who stop contributing and go unresponsive — and the docs note that to date it has never been used.

Eligibility is narrow on purpose: continuous contribution for at least 6 months before inclusion, open source under an OSI-approved license, a regular presence in R&D venues like ethresear.ch and AllCoreDevs. The frame the docs offer is "strictly necessary and existential software required to produce blocks and advance the chain" — which is why MEV-Boost and light clients are excluded.

Funding: the 1% Pledge

The ask is a social norm rather than a protocol rule: projects built on Ethereum donate 1% of their native token supply to Protocol Guild, vesting over four years. Per the 2025 annual report, pledge projects "account for roughly 83% of PG's total funding to date" — 16 named signatories through 2025, including EigenLayer, Ether.fi, Taiko, Puffer, PWN DAO, EthStorage, INTMAX, Shape, Yodl and Ascend.

The 2025 numbers, from the Guild's own reporting:

• All-time inflows: $183m valued at time of donation, of which $7.2m arrived during 2025.
• Outflows: $12.4m distributed to maintainers in 2025, up from $10m in 2024. Median $62k per member over the trailing 12 months.
• 309,183 donations from 6,202 unique donors in 2025. Average donation $23; median donation $0.01 — "we received tens of thousands of micro donations on L2s." A single donation accounted for $4,047,396 of the year's $7.2m.
• Top 2025 donors: Puffer, Starknet, Optimism RPGF, EnterDAO and LayerZero.

The gap between $183m of inflows and $12.4m of annual outflows is the four-year vest doing its job: the mechanism converts lumpy, sentiment-driven token donations into a slow, predictable stream, and no one — donor or member — can accelerate or reverse it.

Where governance does live

Protocol Guild is not ungoverned; its governance is fenced. The mainnet Split contract doubles as the membership registry, and a DAO ratifies changes to it quarterly. That DAO migrated in Q4 2025 from a Moloch V3 deployment (via DAOhaus) to Agora, which also took over ownership of the Split contracts from the Guild's 6/10 Safe multisigs — the report names "reducing multisig dependencies" as a key 2025 goal.

The two proposal types are calibrated to their stakes (architecture docs):

• Split updates — member additions, removals, weight changes — run quarterly and need a simple majority plus 33% quorum, with a 7-day voting period and 2-day grace period.
• Split distributions — pushing already-vested funds out to members — run weekly as "scoped proposals" requiring a simple majority and no quorum.

The asymmetry is the tell. Changing who gets paid is a real decision with a real quorum. Paying out is a formality, so it is built to be unblockable — no quorum means no-shows cannot strand a distribution, the phantom-quorum failure that jams so many token DAOs. Multisig signers are not disclosed publicly and are rotated regularly.

The Cayman entity takes 10% of vested funds to cover legal and operating costs, swapping to stablecoins monthly against a $200k operating reserve and a $100k claims reserve, plus two operations contributors at $12.5k and $10.5k per month. It "aims to receive sufficient funding to cover the above, but not accumulate beyond that" — and the membership sets the percentage, which is precisely the kind of decision that is inside governance scope.

Live tensions

Two things worth watching, both from the Guild's own 2025 report rather than from critics:

Membership shrank for the first time. 186 members at the start of 2025, 184 at the end — 43 added against 45 removed. The report calls this "the first year that Protocol Guild's membership has decreased since its inception," and expects the number to stay "relatively stable around this level." For a mechanism whose legitimacy rests on the registry tracking reality, a decline is evidence the curation is honest, not evidence of decay. But it also means the denominator of a formula that dilutes incumbents has stopped growing.

Time-weighting itself is under review. The ops team "invested substantial effort into exploring different funding distribution methods other than the current time weighting mechanism," culminating in a team impact ranking draft proposal "focused on output and performance over pure tenure/time-weighting." Members at Berlinterop discussed "introducing impact weighting, as well as capping team weights and the number of client slots."

That is the load-bearing question for the whole design. Tenure is objective, cheap to verify, and nearly impossible to game — which is why it can be a formula and therefore why it needs no vote. Impact is what everyone actually wants to pay for, and it is subjective, contestable, and needs a judge. Adopting impact weighting would import exactly the discretionary decision-making that the current mechanism exists to avoid, and the argument for self-curation ("local actors have the most domain knowledge") is also the argument that these particular people could survive it. It is unresolved.

How Caper approaches this

Protocol Guild and Caper answer the same question — how do you stop a treasury from becoming something to fight over? — and they answer it from opposite ends.

Protocol Guild's answer is to make the money unreachable by any vote. It works, and it is the cleanest answer in this wiki. It also only works because of what Protocol Guild is: a funding conduit with no other purpose. There are no decisions to make, so nothing is lost by making decisions impossible. Nobody needs to vote on what to build, because the members are already building it somewhere else and are paid from elsewhere too.

A caper cannot take that route, because a caper is a group that has to actually decide things — fund this, invest in that. Its treasury has to be reachable by proposals, so the fight is live in a way the Guild's never is. Caper's answer is not to remove the vote but to remove the trap: exit is a public method on the contract, so a member who dislikes where the treasury is going leaves with their share without needing anyone's permission — the exit right as a contract function rather than a governance grant.

The two designs rhyme on one point. Both refuse to let a claim be bought. Protocol Guild's share is a function of contributed time and nothing else — no amount of money buys a larger time_weight. Caper's exit share is the same number as its vote weight, (t · v) / (V · T), and v is a soulbound proof-of-vote token: it is minted only by voting and cannot be transferred, so a member who never participates has v = 0 and the formula returns zero regardless of holdings. (Verified against the contract source, contracts/src/caper_dao.rs.)

Two honest qualifications. First, Caper's weight is not purely earned: t — tokens held — is a multiplicative factor, so capital still counts, and a caper's claim is a blend of stake and participation in a way a Guild member's is not. Protocol Guild's allocation is the more radical of the two: it contains no capital term at all. Second, the Guild does not need an exit right because it has no shared pot to be trapped in — each member's claim vests to their own address on a schedule nobody can alter. Caper's exit is doing work that Protocol Guild's architecture makes unnecessary. That is a point for the Guild, within its narrow scope.

The transferable lesson runs the other way, toward every DAO that is neither: if your treasury must be governable, then the question the Guild sidesteps is one you have to answer — what happens to the member who loses the vote?

See also

• DAO contributor compensation — the broader problem of paying people who build a commons
• Retroactive public goods funding — the external-curation model Protocol Guild explicitly rejects
• Gitcoin DAO and the Optimism Collective — the two largest funders that chose councils and rounds instead
• DAO treasury management — what treasuries do when a vote can reach them
• Rage quit and exit rights — the alternative answer to the same question
• How DAOs fail — the failure catalogue this design is built against

References

• Protocol Guild documentation — the primary source for the eligibility framework, registry and contracts
• 1. Membership — eligibility, self-curation, time weight, governance scope
• 2. Smart Contract Architecture — vesting, pass-through wallet, Split v2.1, DAO, multisigs
• 3. Donate — the 1% Pledge and vesting options
• 4. Entity & Operations — Cayman entity, the 10% cut, reserves
• Protocol Guild Annual Report 2025 (29 Jan 2026) — inflows, outflows, membership changes, Agora migration
• Protocol Guild Annual Report 2024 (5 Feb 2025)
• The Protocol Guild Pledge (30 Jan 2024)
• Splits: Protocol Guild funds Ethereum core developers using Splits
• Protocol Guild Dune dashboard — live donation and distribution data