Giveth is a donation platform and a DAO built around one deliberately awkward economic idea: that the person who gives money away should be rewarded for it. Where Gitcoin concentrates on allocating a matching pool well and Protocol Guild on paying a fixed set of maintainers, Giveth takes the demand side: it charges no platform fee on donations, then pays donors back in its own token, and lets token holders stake to decide which projects that reward flows to most generously. The result is a public-goods DAO whose governance surface is unusually wide — three separate voting venues, a working-group budget cycle, and an attestation layer — and one of the longest continuously-running organisations in the sector, with an origin story that predates the word "DAO" being used the way it is now.

From the White Hat Group to a donation platform

Giveth was founded in the second half of 2016 by Griff Green, Jordi Baylina and other members of the Ethereum White Hat Group — the same people who had just spent the summer working through the fallout of The DAO. Their first output was infrastructure rather than a product: the MiniMe token contract, a Vault contract, and a MilestoneTracker, all open-sourced (Giveth history). MiniMe — an ERC-20 that keeps a historical record of every holder's balance, so a snapshot at any past block can be read on demand — became a standard building block for token voting across the ecosystem.

The group's most-cited moment came on 19 July 2017, during the Parity multisig incident. By Giveth's own account, its developers used the same vulnerability an attacker was exploiting to drain 500 at-risk wallets into a secure account before the attacker could reach them, rescuing 47,000 ETH in the first pass and over $208 million of crypto in total, then rebuilt secure wallets and returned the funds to their owners by 31 July, declining donations for the work. The same year the team declared itself a Decentralized Altruistic Community and reorganised on holacratic lines — circles, weekly governance meetings — a structure that survives today as the DAO's Working Groups and Chapters.

The consumer product arrived in two waves. The original 2017 DApp eventually shipped as Giveth TRACE in July 2021; a rewritten platform launched as Giveth.io in March 2021 with peer-to-peer donations on mainnet and xDai (now Gnosis Chain). The token economy — announced by Green at The DAOist in Paris that July — followed as the GIVeconomy.

The GIVeconomy: GIV, GIVstream and GIVbacks

1 billion GIV were minted at the launch of the GIVeconomy. Only 100 million were liquid immediately; the remaining 900 million were committed to the GIVstream, a continuous release running until 23 December 2026 (GIVeconomy docs). Anyone who claims GIV from any part of the economy also opens a stream of further GIV that drips out over the remaining term — a vesting design that ties reward size to how early you participated rather than to a cliff schedule, and one that is now within months of fully unlocking.

GIVbacks is the demand-side mechanism. Donate at least $5 in an eligible token to a verified project and Giveth returns 50% to 80% of the USD value of the donation in GIV, with the exact percentage set by the receiving project's GIVpower rank. Rounds last two weeks and each round has 1 million GIV available; if donations exceed that budget, every donor is scaled down proportionally rather than the round closing early (GIVbacks docs). Giveth itself takes no platform fee on donations, so GIVbacks is the platform's entire economic pitch to a donor: give, and get a claim on the platform's governance back.

GIVpower: paying people to curate

The supply-side counterpart is GIVpower, and it is the part of Giveth most worth studying as a governance mechanism. Staking GIV in the GIVfarm mints GIVpower 1:1. Stakers may then lock for between 1 and 26 rounds of two weeks each — up to a year — and locking multiplies both the staking APR and the GIVpower by √(1 + N), where N is the number of rounds locked. Ten rounds gives a multiplier of about 3.32; the maximum 26-round lock gives about 5.20 (GIVpower docs).

That is a square-root time-lock curve, and the contrast with the linear-decay curve of vote-escrow tokenomics is the interesting part: a square root pays most of its bonus in the first few rounds and flattens after, so it buys commitment without making the longest lock overwhelmingly dominant. Holders spend GIVpower by boosting up to 20 projects at a time. Ranks are recomputed each GIVbacks round from the average GIVpower behind a project across the previous round, and the rank sets the donor reward: the top-ranked project yields up to 80% GIVbacks, an unranked verified project 50%.

So GIVpower is a curation market in which the curated good is a discount on somebody else's donation. It gives stakers a reason to research projects that has nothing to do with a proposal vote, which is a rare answer to voter apathy: the work is continuous, the reward is continuous, and no ballot has to be scheduled for any of it to happen.

Three governance venues, one DAO

Giveth splits decisions across three systems rather than routing everything through one contract (governance process).

The GIVgarden runs on conviction voting via 1hive's Gardens framework on Gnosis Chain, and is where treasury requests land. Members wrap GIV into gGIV and stake it behind proposals; conviction accrues the longer a stake sits. Its published parameters are a 30-day conviction growth, a 2.50% spending limit per proposal, 5% minimum conviction, a 5,000 GIV action deposit and a 10,000 GIV challenge deposit, with a 7-day settlement period. Non-financial questions use Tao voting — binary, delegable, with quiet-ending protection against last-minute flips — at 90% support required. Challenged proposals escalate to Celeste, a decentralised court, as the final backstop (GIVgarden docs).

rGIV is a non-financial reputation token held by DAO members, voted on an Aragon deployment with plain yes/no mechanics and a five-day window. Only rGIV holders may propose or vote there, which keeps operational decisions inside the contributor body rather than the token market. That venue looks dormant in practice: the client URL the governance docs publish for it, optimism.aragon.blossom.software, returns NXDOMAIN as of 2 August 2026, so the live governance load is carried by the GIVgarden and Snapshot.

Snapshot carries community-wide signalling and budget ratification in the giv.eth space, where GIV held and GIV staked in GIVpower both count toward voting power. As of 2 August 2026 the space has run 73 proposals for 315 followers, with a five-day voting period (snapshot.org/#/giv.eth). Two things stand out in the live data. First, a documentation gap: the governance docs state the quorum is 1 million GIV, while the space's own Snapshot configuration returns a default quorum of 10 million — a tenfold difference between what the docs tell a prospective voter and what the voting client enforces. Second, turnout: recent proposals close with roughly 10 to 29 voting addresses apiece, including a July 2026 request for 8.5M GIV to extend GIVpower rewards to February 2027 that drew ten. Voting power is in the hundreds of millions and quorum is never the binding constraint; the number of distinct humans is. That is the same concentration pattern documented across token-weighted DAOs, and it holds even where the token is explicitly earned by giving money away.

Working Groups, Seasons and the Advice Process

Operationally the DAO is split into Working Groups — currently Dapp, GIVeconomy, DAO Ops, Quadratic Funding and Fundraising — each owning its own budget, goals and contributors, plus skill-based Chapters (Development, Design, Comms & Marketing, Product, devOps) that cut across them and hold no budget of their own (DAO structure). Working Groups run in Seasons of three or six months; at each boundary a lead posts what the group achieved and what it plans next, with goals framed at three ambition levels — SHRINK, SUSTAIN and GROW — and GIV holders vote on which groups to fund. It is a clean, legible instance of the working-group pattern, and the three-scope framing is a neat trick for budget votes: the electorate picks an ambition level rather than haggling over a single number.

Before anything reaches a vote it must clear the Advice Process: every legitimate proposal starts on the Giveth forum and must sit open for feedback for a minimum of five days. A proposal may shorten or skip that window in a financial emergency, but only with a written disclaimer explaining why — and the docs warn plainly that doing so lowers its chance of passing. Conduct is governed by a signed Community Covenant, which is also the reference document Celeste rules against when a GIVgarden proposal is challenged.

Quadratic funding, and running GG24's OSS rounds

Giveth has run quadratic-funding rounds on its own platform since 2023, matching donations to verified projects out of a pool held by the donation.eth multisig, a 7-of-15 signer set managed by core contributors (matching-pool docs). Matching-fund distributions are ratified on Snapshot after the fact rather than executed automatically, which puts a human ratification step between the QF calculation and the payout.

In October 2025 Giveth operated the open-source software rounds of Gitcoin's GG24 — the first round of "Gitcoin 3.0", which spread roughly $1.8M across six domains using six different allocation mechanisms at once. Giveth ran two of them (Developer Tooling & Infrastructure, and Interop Standards, Infra & Analytics) against a $300,000 matching pool, drawing $36,657 in crowdfunding from 1,286 unique donors across 3,042 donations (GG24 OSS QF retrospective). The grantee count depends which Gitcoin document you read: the results thread lists 55 projects in the tooling round and 23 in the interop round — 78 in total — while the retrospective counts 64 funded projects. Every other figure agrees exactly across the two: $29,739 + $6,918 in donations, 2,361 + 681 individual donations, and 200,000 + 100,000 USDC of matching.

The retrospective is candid about what those numbers mean. The crowdfunding-to-matching ratio came in at 12.22%, below GG23's 15.88% and GG22's 19.43% — every dollar of matching pulled in fewer donor dollars than the round before. Donor count beat three comparable 2025 Giveth rounds, but donations per donor fell to 2.36, and the round's much tighter eligibility (64 projects, against 235–496 historically) is named as a cause alongside a general slowdown in public-goods funding. That slowdown is the same one visible in the pause of Optimism's Retro Funding in January 2026, and it is the backdrop against which every funder in this section is now operating.

DeVouch: attestations as a legitimacy layer

DeVouch is Giveth's answer to the question every grants platform eventually hits — not "is this donor a real person" but "is this project what it claims to be". It is built on the Ethereum Attestation Service: recognised Attester Groups, defined as sets of addresses holding the same attestation from the same issuing authority, publish vouches and flags against projects, and DeVouch indexes those signals across Gitcoin, Giveth and Optimism Retro Funding in one place (DeVouch docs). Anyone may attest, but only groups on the indexed list are surfaced — Optimism's RetroPGF badgeholder schema is the worked example, and new groups are added through the public repository.

It is a deliberately different shape from the identity-side approaches catalogued under Sybil resistance: rather than scoring the wallet that donates, DeVouch scores the grantee, and it does so with reusable, portable, public attestations instead of a platform-private review queue. For a funder deciding whether a project deserves a matching multiplier, that is the cheaper half of the problem to attack.

Where Giveth sits

Among the public-goods funders in this directory, Giveth occupies the retail-donor end. Gitcoin is a mechanism laboratory that allocates matching pools; Protocol Guild is a time-weighted split contract for a known set of maintainers; MolochDAO is a small, ragequit-protected grant committee. Giveth is the one trying to make the ordinary donation economically attractive, and to fund itself by governing who benefits from that subsidy.

The design deserves its scrutiny in both directions. It is a genuine experiment in paying for attention to public goods continuously rather than in scheduled votes, and GIVpower is one of the better-specified curation mechanisms running in production. But the honest reading of its own numbers is that it has not escaped the pattern: participation concentrates, turnout is a handful of addresses, and the GIVstream that funded the whole reward economy runs out in December 2026. What Giveth does next — with the stream exhausted and the crowdfunding ratio falling across the sector — is the more informative thing to watch than anything in its documentation.