What it is
Quadratic funding is a method for allocating a pool of matching funds across public-goods projects based on crowdfunding patterns. Instead of matching donations dollar-for-dollar, it rewards the number of distinct contributors far more than the size of any single contribution — so a project with 100 people each giving $1 gets matched much more heavily than a project that received one $100 donation, even though both raised the same amount directly.
How it works
- A round is set up with many eligible projects and one shared matching pool, usually funded by a sponsor, protocol treasury, or donations.
- Individual donors contribute any amount they want to any projects they choose during the round.
- For each project, the system takes the square root of each individual contribution and sums those square roots, then squares that sum — this "sum of square roots, then squared" calculation is the quadratic part.
- Because square-rooting a large number gives proportionally less than square-rooting many small numbers that add to the same total, this formula produces a much higher score for broad-based support (many small donations) than for concentrated support (few large donations) of equal total value.
- Each project's calculated score, minus what was actually donated, roughly represents its "ideal" matching amount reflecting the strength of its broad-based demand.
- Because the matching pool is limited, every project's ideal match is typically scaled down proportionally so the total payout fits within the available pool.
- At the end of the round, the matching pool is distributed according to these scaled amounts, on top of the direct donations each project already received.
- Results and calculations are typically published so anyone can verify the payouts.
Why designers use it
- Surfaces genuine grassroots demand for public goods rather than letting whoever has the deepest pockets decide funding priorities.
- Lets a sponsor's matching budget go further in expressing collective community preference instead of the sponsor picking winners directly.
- Encourages projects to build a genuine base of many small supporters rather than courting one wealthy patron.
- Provides a transparent, formulaic, and auditable way to fund public goods without a central committee making the final call.
Failure modes
- Sybil attacks: since the formula rewards contributor count over size, an attacker can split themselves into many wallets each donating a small amount to inflate their own project's match — the core exploit target of every QF round.
- Collusion rings: groups of real people coordinating to donate small amounts to each other's projects can extract disproportionate matching funds without genuine broad public support.
- Whale-disguised-as-many: a well-funded backer distributing funds across many wallets to donate "small" amounts can mimic organic grassroots support.
- Pool exhaustion by volume: if far more legitimate demand shows up than the matching pool can support, everyone's payout gets diluted, sometimes making participation feel not worth the effort.
- Identity verification friction: strong sybil resistance usually requires proof-of-personhood or reputation systems, which add UX friction, cost, or centralization risk to what's otherwise a purely algorithmic mechanism.
What to check before using it
- Decide what sybil-resistance layer you'll pair with QF (proof of personhood, social graph analysis, stake-based identity) since the formula is only fair if one-person-one-identity roughly holds.
- Model how collusion detection and result review will work — most real deployments retroactively adjust or flag suspicious donation clusters.
- Size the matching pool relative to expected donor turnout so payouts remain meaningful rather than diluted to near-zero.
- Decide on round cadence and eligibility criteria for projects, since QF's fairness properties assume a reasonably open, contestable set of participants.
- Plan for transparent publication of contribution data and calculations, since the mechanism's legitimacy depends on being independently auditable.