Balancing Power in Decentralized Governance : Quadratic Voting under Imperfect Information

Blockchains hold the potential to revolutionize our economic systems, but their success hinges on effective self-governance. One crucial question is how these decentralized systems should gather user input — a hotly debated topic. Though most blockchains rely on a form of linear stake-weighted voting (LV), this mechanism can be suboptimal (Tsoukalas and Falk 2020). Quadratic Voting (QV) has been proposed as a possible solution — it can restore optimality when users have perfect information on the proposals they are voting for (Lalley and Weyl 2018b, Buterin et al. 2019). But what if they don't? We develop a game-theoretic model of QV-based governance and show that optimality breaks under imperfect information: QV can over- or under-perform LV. In brief, this is because the squaring mechanism it relies on can make it costlier for better-informed voters to convey their private information. On the other hand, we find that both LV and QV can restore optimality asymptotically (as the number of voters tends to infinity), at which point, they become equivalent. These results bring to light an intuitive, yet often overlooked dimension in decentralized governance design: the need to boost voter turnout — a task that many blockchains struggle with (Falk and Tsoukalas 2023)