Yesterday, I discussed a phenomenon with a friend: the open-source library he maintains has doubled its PR volume in the past three months, but the time he spends on reviews has actually decreased.

It's not laziness; he simply can't keep up, and many PRs are obviously AI-generated—logic can check out, but something feels off.
He said something that left a deep impression on me: “Now, people submitting code don’t have to bear the consequences, but those merging the code do.”
This statement highlights the core issue: in open-source collaboration, the risks are asymmetrical between contributors and reviewers.
If a PR submitter writes faulty code, the loss is the project's reputation and the time spent fixing bugs later; even if reviewers work tirelessly as volunteers, they get no compensation.
This asymmetry is amplified in the AI coding era—generating code is too easy, but verifying it remains very difficult.
MergeProof aims to address this imbalance.
It allows contributors to stake a sum of money when submitting a PR, signaling “I believe this code is fine.” If reviewers find vulnerabilities, the staked money becomes a reward; if no issues are found, the money is returned. It’s that simple.
But behind this simplicity is a fundamental shift: code review transforms from “a reputation-based altruistic act” into “an incentive-based economic activity.”
You no longer rely on others’ goodwill to review your code; you just need to stake on your own, and naturally, people will verify for profit.
The more I think about it, the more I believe this could be the foundational layer missing in open-source collaboration.
Everyone talks about Web3 bringing economic incentives into various scenarios, but MergeProof’s approach is very precise—it doesn’t invent a new process, it just adds a game-theoretic mechanism to the existing one.
Contributors can demonstrate their confidence, reviewers can monetize their time, and project teams get higher-quality code—all three parties benefit.
Moreover, it’s especially suitable for this explosion of AI coding. As code volume begins to grow exponentially, relying on “volunteer spirit” alone can’t sustain the review process. Economic incentives might be the only scalable solution.
I’ve been watching it for a while, and I feel it could gradually change the way software collaboration works.
At least, it made me start to wonder: if I truly have confidence in my code, would I dare to stake?