Vitalik Buterin backs Gitcoin’s Deep Funding after GG23 predictive challenge, where 11 models competed to forecast $1.5M grant allocations. Best score: 0.016.
Gitcoin just wrapped a major round of Deep Funding. Ethereum co-founder Vitalik Buterin took notice and shared his thoughts publicly on X.
In a post on X, Vitalik Buterin confirmed Deep Funding had completed a significant new round. He called out the GG23 Predictive Funding Challenge as a project worth refining, including a prediction market version he wants to see developed further. His comments landed right as the retrospective for the challenge went public on the Gitcoin governance forum.
You might also like: Bitcoin Season Dominates Crypto Market As Analysts Watch For Next Altcoin Run
The 36-Hour Race That Changed Grant Funding
Gitcoin ran the predictive challenge during GG23. Data scientists had just 36 hours to predict how $1.5 million in grants would be split across projects, before the round even opened to the public. That narrow window, according to the GG23 retrospective on the Gitcoin governance forum, was one limitation that held back wider participation.
Eleven valid models were submitted. Seven of those teams wrote up full explanations of their methods. The competition offered a $10,000 prize pool, split equally between Gitcoin and the Ethereum Foundation’s Unblocking Pilots team.
First place went to David Gasquez. Error score: 0.016. Second went to Omniacs DAO at 0.0173, and Limonada took third at 0.018. A model with no useful features would produce an error score around 0.04, making the top submission’s performance more than twice as accurate as random guessing.
Must read: Arthur Hayes Warns Bitcoin Rally May Only Be a Temporary Bounce
Buterin Sees Promise But Flags a Design Problem
As Buterin posted on X, his main advice to Gitcoin contributor @devanshmehta was to keep refining the experiment but rethink how the design aligns with what he called “chaotic era needs.” He referenced his own recent thread on democratic systems when framing the issue.
Buterin said Deep Funding already scores well on being meritocratic rather than over-egalitarian. He also said it handles AI involvement in a way that keeps human agency intact. But there’s a catch.
Looking at how the system is built right now, it gives off what he called a “stable era” vibe. The structure assumes large-scale social agreement to pump money into a single mechanism. That, Buterin indicated, does not fit the current moment in crypto governance.
Why Predictive Models Could Reshape Grant Allocations
The GG23 retrospective laid out several reasons why predictive modeling rounds should become standard practice for any public grant program. Winning models in this round relied heavily on XGBoost algorithms. Key features they used included the number of contributors, the past amount per contributor, and the sentiment score for each project.
The idea of “shadow rounds” came up in the forum post. That is where a winning model from a past round gets queried to estimate how much funding a new project would receive, even if that project never entered the actual grant round. The lower the error score, the more reliable the shadow allocation becomes.
You might also like: Chainlink Powers Visa, ANZ, Fidelity in Cross-Border Settlement
Prediction markets were also proposed alongside the modeling competition. Traders would wager on how much funding each project receives, with markets resolving once final allocations are announced. That pulls in a much larger audience of traders, beyond just existing grant donors.
What Comes After GG23
The retrospective, posted by thedevanshmehta on the Gitcoin governance forum, noted that future rounds need a longer submission window, ideally a full week. It also flagged that some design questions remain open, including how round operators choose between competing allocation lists without favoring known associates.
Buterin’s public comments add weight to those questions. He tied his feedback directly to broader thinking on democratic processes, linking to a post on Firefly where he had discussed the topic in more depth. His view was that Deep Funding has the right instincts but needs structural adjustments for a less predictable environment.
Gitcoin’s stated business goal from all of this is to become the platform where allocation models are submitted and queried for a fee. Any funder wanting a lightweight method to distribute money could pick from a ranked list of models, based on how well each performed in past rounds.
Also read: ICE Invests in OKX at $25B Valuation to Expand Tokenized Stock Trading
GG25 rounds are the next target. Gitcoin expects more funders to adopt predictive challenges alongside their standard grant processes going forward.
