Decentralized science funding is the family of mechanisms DeSci uses to route capital to research without a central grant committee: community-weighted matching, pay-for-demonstrated-impact, token-holder-curated grants, and treasuries that fill continuously from token markets rather than episodic fundraising.

Why science funding is being rethought

Traditional grant funding concentrates decisions in a small number of agencies and committees, with long cycles and heavy application overhead. The pressure became acute after the 2025 wave of U.S. federal research-funding cuts: a JAMA Internal Medicine study (Patel, Liu & Jena, JAMA Intern Med. 2026;186(1):126–128) found that NIH grant terminations disrupted 383 active clinical trials — about 1 in 30 — affecting more than 74,000 enrolled participants. Against that backdrop health-research commentators began treating decentralized funding, governance, and data-sharing as credible alternative infrastructure rather than a Web3 curiosity. Roughly 50 active DeSci initiatives now span funding DAOs, publishing, healthcare data, and IP management.

Quadratic funding

Quadratic funding amplifies a matching pool by the breadth of community support: many small donations attract more matching than one whale donating the same total, because the match scales with the square of the sum of square roots of contributions. Gitcoin made it the default mechanism for open-source and public-goods funding and has run dedicated DeSci rounds distributing community-matched grants to research projects. The mechanism is set out in full on quadratic voting and funding.

The most recent large round shows both the scale and the strain. Gitcoin Grants 24 (GG24) distributed over $1.8 million across six thematic domains, with donation rounds running 14–28 October 2025. Of that, $1.175 million was Gitcoin's own matching and roughly $632,500 came from external partners — the first Gitcoin round in which more than a third of the matching capital originated outside the core DAO treasury. The two open-source quadratic-funding domains, operated with Giveth, drew $36,657 in direct donations amplified by $300,000 in matching to 78 projects from roughly 1,300 unique donors (results thread).

GG24 is also notable for refusing to be a single-mechanism round: it combined quadratic funding with conviction voting, retroactive funding, MACI private voting, Deep Funding, and peer-reviewed hypercerts in one cycle. For a DeSci funder the practical reading is that no single allocation rule is treated as settled — matching, judged impact, and private ballots are now run side by side and compared.

Retroactive funding

Retroactive public-goods funding inverts the grant: instead of predicting which proposals will succeed, it rewards work that has already proven valuable — on the theory that it is easier to agree on what was useful than on what will be. The model was popularized by Optimism's Retro Funding rounds, which have distributed hundreds of millions of OP to infrastructure and tooling, and DeSci programs borrow it for reproducibility work and datasets whose value only shows after publication. The full design, its judging problem, and its results are covered on retroactive public goods funding.

The canonical implementation is currently paused. Announcing Season 9 on 8 January 2026, the Optimism Foundation stated that “the Retro Funding program will not run for at least the next 12 months”, and flagged a possible re-allocation of the ~775M OP reserved for it. Any DeSci programme designing around retroactive funding should treat Optimism as a source of method rather than as a live matching partner — see Optimism Collective.

Token-curated grants and IP-NFT deal flow

DeSci DAOs put funding decisions to their token holders: proposals are sourced and reviewed (typically by expert working groups), then approved by token vote, with the funded intellectual property tokenized as an IP-NFT held by the DAO — a structure pioneered by Molecule (docs). VitaDAO has deployed $4.7M across 31 research projects this way, and vertical DAOs replicate the model per disease area — HairDAO for hair loss, CryoDAO for cryopreservation research.

Where that expert review is scored and published, it becomes a gate in its own right rather than advice to voters. VitaDAO runs a "senior review" in which a panel scores a proposal out of 5 across seven categories before it reaches a token vote: VDP-164, a $500,000 loan to a portfolio company, drew an overall conviction score of 2.3 against an average of 3.7 for previously funded projects, and has not appeared on the DAO's Snapshot space since. A low score does not reject a proposal under any written rule – it strands one, which is a different accountability problem from the one token voting is usually accused of.

Launchpads and curve-funded treasuries

The newest layer skips grant rounds entirely: launchpads such as Bio Protocol curate new bioDAOs and fund them through token launches, so the research treasury fills from an open token market and keeps filling as the token trades — continuous funding priced by the crowd instead of a committee.

What the 2026 retrenchment means for research funders

Two of the sector's largest public-goods funders moved in the same direction within a year, and the pattern matters more than either event. Optimism paused its programme outright; Gitcoin kept running but sourced more than a third of GG24's matching pool externally rather than from its own treasury. Both are symptoms of the same structural fact: a matching pool funded from a protocol treasury is a discretionary expense, and discretionary expenses are the first thing a DAO cuts.

That is a direct argument for the funding models the rest of this page describes. IP-NFT deal flow gives a DeSci DAO a claim on the assets it funds rather than a donation receipt; a curve-funded treasury fills from trading rather than from an annual allocation vote. Neither is obviously better than a well-run matching round — but neither can be switched off by a single governance post, which is what a research group planning multi-year work now has to weigh.

How Caper approaches this

Caper is built around that last model. Launching a caper puts a research community's treasury on a bonding curve: anyone can buy in at a price set by the curve, and part of every trade accrues to the caper's own treasury, so funding is continuous rather than a one-shot round. The community then directs the treasury through on-chain proposals — PAYOUT pays a lab, supplier, or researcher directly; INVEST lets one research caper back another. Because members who have voted keep an exit right at all times, backers are never locked into an agenda they no longer support. That exit pays out the member's canonical vote weight (t·v)/(V·T) — the same number that sets their voting power — so the share tracks participation as well as stake, rather than being a flat pro-rata claim on tokens alone.