New study reveals 58.8% voting bias from author choices in DAOs

A new paper from Stefano Balietti, Pietro Saggese, and Markus Strohmaier (arXiv:2607.09435) reveals significant voting biases in Decentralized Autonomous Organization (DAO) governance. By analyzing proposal-choice level data, the researchers found that author-selected choices receive a striking 58.8% increase in voting-power share compared to non-author choices. Approval-oriented choices show a 27.1% advantage, while first-listed choices benefit from a 7.7% position effect. These biases persist even after subtracting the author's own voting power, indicating that interface design and social signals—not just token weight—drive concentration in DAO voting.

The authors describe these as systematic associations rather than proven causal distortions, but the implications are clear: DAO governance is not as neutral as its token-weighted voting suggests. Ordering of choices, author signals, and vote visibility should be treated as institutional design choices, not trivial implementation details. This shifts attention from proposal outcomes to the ballot experience itself. For the growing Web3 ecosystem—where DAOs manage billions in assets—these biases can undermine legitimacy and fairness. Designers may need to randomize order, hide author signals, or rethink approval voting to reduce distortions.