Voting Biases in Decentralized Autonomous Organization (DAO) Governance
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Computer Science. The day in computer science: software, systems and theory on arXiv.
Tom: We're looking at the paper: "Voting Biases in Decentralized Autonomous Organization (DAO) Governance".
Jane: Here's the gist of the paper. Decentralized Autonomous Organizations (DAOs) use token-weighted voting to allocate resources, set protocol rules, and legitimate collective decisions <ref:2607.09435#pg2> Yet support in DAO voting is strikingly concentrated <ref:2607.09435#pg2>.
Tom: Key Findings on Voting Biases The study finds that author-selected choices show the strongest and most robust association with votingpower share, with a 58.8% increase relative to non-author choices <ref:2607.09435#pg2>.
Jane: Approval-oriented choices retain a positive but slightly less consistent advantage (27.1%) <ref:2607.09435#pg2>.
Tom: First-listed choices also attract systematically higher shares, consistent with position and order effects (7.7%) <ref:2607.09435#pg2>.
Title and authors: Tom: We're talking about "Voting Biases in Decentralized Autonomous Organization (DAO) Governance," and the authors are Stefano Balietti, Pietro Saggese, and Markus Strohmaier.
Jane: It sounds like they're looking at how the mechanics of voting inside a DAO—which uses token-weighted voting to make decisions—can lead to a surprising concentration of support.
Lu: The paper sets up the problem by asking what happens inside that ballot that produces this concentration, and they study it at the proposal-choice level.
Meng: So instead of just looking at the final decision, they're zooming in on how individual choices influence the outcome based on presentation and author signals.
Tom: The core finding they present is that there are three specific features—approval stance, choice list position, and author selection—that are associated with voting power share.
Jane: They quantify this: author-selected choices have a fifty-eight point eight percent increase in voting power compared to non-author choices.
Lu: And they also found that approval stances have a positive but smaller advantage, around twenty seven point one percent.
Meng: It seems like the study is showing that while people's actual approval matters, the way things are set up—who wrote it and where it sits on the list—is a bigger driver of support than you might expect.
Tom: And they also found that first-listed choices attract systematically higher shares, about a seven point seven percent increase because of position effects.
Jane: So when you put those three things together—the author's choice, the approval signal, and the list order—they are all connected in a way that explains why support gets stuck in certain places.
Lu: They go on to decompose these biases into position bias, approval bias, and author bias to see how much each channel is responsible for that concentration.
Meng: That decomposition shows they aren't just looking at one factor; it’s a combination of everything working together to create the pattern they are seeing.
Tom: The average marginal effects calculation is pretty telling: the author-selected choice gains fifty-eight point eight percentage points on average, which is more than double what approval choices get.
Jane: That comparison really hammers home how much weight that author signal carries in this system compared to other factors.
The paper's summary: Tom: To summarize the study of "Voting Biases in Decentralized Autonomous Organization (DAO) Governance," they looked at how voting power share for a choice is affected by whether it expresses an approval stance, where it appears in the list, and if it was selected by the author.
Jane: The main point is that these three things—the author signal, the position in line, and the approval quality—are all statistically tied to how much voting power a specific choice captures within a DAO vote.
Lu: They found that author-selected choices have the strongest association with voting power share at fifty-eight point eight percent more than non-author choices.
Meng: That means if you want to influence the outcome, making sure your choice is selected by the proposal author is probably your biggest lever, based on this data.
Tom: They also found approval-oriented choices have a positive but less consistent advantage of twenty seven point one percent.
Jane: So even if you don't control the author signal, having a good approval stance still helps, but it’s not as reliably strong as author selection.
Lu: Additionally, first-listed choices attract systematically higher shares by about seven point seven percent due to position bias in the ballot order.
Meng: So it’s a layered effect: you have the author's signal, plus where you are placed on the list and how agreeable your choice is.
The paper's improvements: Tom: The authors of "Voting Biases in Decentralized Autonomous Organization (DAO) Governance" suggest several practical improvements for how these systems are designed.
Jane: They want to move beyond just observing these biases and actually build mechanisms that make them transparent.
Lu: One major suggestion is to implement bias-aware governance design, which means making things like choice order, author-choice visibility, and author-vote visibility documented and auditable instead of treating them as neutral interface details.
Meng: So they are pushing for the system itself to expose the institutional design choices that influence voting power accumulation so people can see it happening.
Tom: Another point is about stance classification: they suggest developing a stance classification procedure, which they plan to triangulate with external model and human evaluation checks to robustly classify stances into Approve, Reject, Abstain, or Other across different proposal formats.
Jane: That sounds like a way to make sure that the qualitative findings about approval biases hold up even when we look at different ways of labeling those choices.
Lu: They also suggest using regression models to separate the conditional associations between voting power share and position bias, approval bias, and author bias to really understand which specific channel is causing that non-random concentration of support.
Conclusion: Tom: So wrapping up the "Voting Biases in Decentralized Autonomous Organization (DAO) Governance" paper, the implication is a call for designers to be much more transparent about these voting mechanics.
Jane: They are stressing that voters and observers need clearer information about who created a proposal, what option the author selected, and how much of the final score comes from that author's own voting power.
Lu: The paper concludes that choice order, author-choice visibility, and author-vote visibility should be documented and made auditable rather than treated as neutral interface details in DAO governance.
Meng: From a production standpoint, it means designing interfaces where these factors are explicit data points so we can measure how much influence each one actually has on the final vote share.
Tom: It’s about making the institutional design choices visible so we aren't just accepting the outcome without knowing what caused it.
Jane: So, to wrap up, this paper shows that author signal is a heavy driver, and visibility into those signals needs to be part of good governance design for DAOs.
Lu: The big picture here is understanding how voting power accumulates in decentralized systems through these specific presentation and social cues.
Meng: I'm curious what the next piece on arXiv will be, because we need to keep tracking this kind of structural analysis for real-world deployment challenges.
Stefano Balietti, Pietro Saggese, Markus Strohmaier
IMT School for Advanced Studies Lucca · Complexity Science Hub Vienna · University of Mannheim · GESIS - Leibniz Institute for the Social Sciences
cs.CY, cs.HC, econ.GN, q-fin.EC
Submitted: 2026-07-10
Updated: 2026-11-03
License: http://creativecommons.org/licenses/by/4.0/
Importance score: 92/100
The gist: Decentralized Autonomous Organizations (DAOs) use token-weighted voting to allocate resources, set protocol rules…
Key concepts
- Author Selection Bias
- This bias occurs when the choice made by the person who created or suggested a proposal receives disproportionately more voting power than other options. The study found this effect is very strong, showing author-selected choices gain significantly more voting share.
- Position Bias
- This refers to the tendency for voters to favor choices based on their order in which they are presented. First-listed options consistently attract a higher share of voting power than subsequent options, demonstrating an effect related to list ordering.
- Approval Bias
- This bias measures the tendency for voters to favor choices that align with their approval orientation. While this effect is positive, it is less consistent and weaker than the biases related to author selection or choice position in DAO voting.
Terminology
Summary
Decentralized Autonomous Organizations (DAOs) use token-weighted voting to allocate resources, set protocol rules, and legitimate collective decisions <ref:2607. p. 0 Voting Biases in Decentralized Autonomous
Organization (DAO) Governance
Stefano Balietti1
, Pietro Saggese2,3
, and Markus Strohmaier3,4,5
1 Private Sector
2
IMT School for Advanced Studies Lucca
3 Complexity Science Hub Vienna
4 University of Mannheim
5 GESIS - Leibniz Institute for the Social Sciences
Abstract. Decentralized Autonomous Organizations (DAOs) use token-weighted voting to allocate resources, set protocol rules, and legitimate collective decisions. Decentralized Autonomous Organizations (DAOs) use token-weighted voting to allocate resources, set protocol rules, and legitimate collective decisions. Yet, support in DAO
voting is strikingly concentrated.Where the paper says it
Where the paper says it
Key Findings on Voting Biases
The study finds that author-selected choices show the strongest and most robust association with votingpower share, with a 58.8% increase relative to non-author choices <ref:2607. p. 0 We find
that (i) author-selected choices show the strongest and most robust association with votingpower share, with a 58. 8% increase relative to non-author choices; (ii) approval-oriented choices
retain a positive but slightly less consistent advantage (27. We find
that (i) author-selected choices show the strongest and most robust association with votingpower share, with a 58. 8% increase relative to non-author choices; (ii) approval-oriented choices
retain a positive but slightly less consistent advantage (27. In DAO governance, ordering,
author signals, and vote visibility should be treated as institutional design choices, not neutral
implementation details. JEL Codes: D71, D72, G23, G34, O33
Keywords: Decentralized Autonomous Organization, DAO, Governance, Bias, Blockchain,
Voting, Collective decision-making, Decentralized Finance, DeFi
arXiv:2607. 8% increase relative to non-author choices; (ii) approval-oriented choices
retain a positive but slightly less consistent advantage (27. 1%); and (iii) first-listed choices also
attract systematically higher shares, consistent with position and order effects (7. In DAO governance, ordering,
author signals, and vote visibility should be treated as institutional design choices, not neutral
implementation details. JEL Codes: D71, D72, G23, G34, O33
Keywords: Decentralized Autonomous Organization, DAO, Governance, Bias, Blockchain,
Voting, Collective decision-making, Decentralized Finance, DeFi
arXiv:2607. 1%); and (iii) first-listed choices also
attract systematically higher shares, consistent with position and order effects (7. Results
are robust across several specifications, which include subtracting an author’s own voting power
from computations.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Decomposition of Bias Effects
The paper investigates the extent to which high approval rates and one-sided outcomes observed in Snapshot can be explained by three biases: position, capturing the stance of a choice, and author selection <ref:2607.09435#pg2>. The authors define these as position bias, approval bias, and author bias <ref:2607. p. 2 , approval
bias. Third, choices selected by proposal authors may reflect author voting power, an informative
author signal, or social influence from visible early voting and reputational cues [48,52,25], a
pattern we refer to as author bias. 1: Share and excess of choice-level voting power for the three indicators under
study. Outcome: mean voting-power share for choices that are first-listed, approval-oriented,
or selected by the proposal author.Where the paper says it
Where the paper says it
Empirical Modeling and Validation
The researchers use fractional-logit models with a quasibinomial logit link to estimate the associations between voting-power share and the three biases <ref:2607.09435#pg2>. The average marginal effects (AMEs) show that an author-selected choice gains 58. p. 0 We study DAOs’ governance at the proposal-choice level, linking each choice’s votingpower share to three observable features: whether it expresses an approval-oriented stance,
where it appears in the choice list, and whether it is selected by the proposal author. We find
that (i) author-selected choices show the strongest and most robust association with votingpower share, with a 58. We find that author-selected choices gain 58. 8 percentage points of voting-power share on
average, more than double the 27. 8 percentage points of voting-power share on
average, more than double the 27. 1% advantage of approving choices and nearly eight times the
7. 8% increase relative to non-author choices; (ii) approval-oriented choices
retain a positive but slightly less consistent advantage (27. 1%); and (iii) first-listed choices also
attract systematically higher shares, consistent with position and order effects (7. Notably, our findings are valid across several alternative specifications; even after
recomputing estimations by removing author’s own voting power, our headline results do not
change substantially. Understanding these associations is crucial to shedding light on the quality and robustness of
collective decision-making in DAOs.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Implications for Governance Design
The findings point to a practical transparency issue in DAO governance, suggesting that voters and observers need clearer information about who created a proposal, which option the author selected, and how much of the final score comes from the author’s own voting power <ref:2607.09435#pg2>. The paper concludes that choice order, author-choice visibility, and author-vote visibility should be documented and made auditable rather than treated as neutral interface details <ref:2607. p. 0 In DAO governance, ordering,
author signals, and vote visibility should be treated as institutional design choices, not neutral
implementation details. JEL Codes: D71, D72, G23, G34, O33
Keywords: Decentralized Autonomous Organization, DAO, Governance, Bias, Blockchain,
Voting, Collective decision-making, Decentralized Finance, DeFi
arXiv:2607. In DAO governance, ordering,
author signals, and vote visibility should be treated as institutional design choices, not neutral
implementation details. JEL Codes: D71, D72, G23, G34, O33
Keywords: Decentralized Autonomous Organization, DAO, Governance, Bias, Blockchain,
Voting, Collective decision-making, Decentralized Finance, DeFi
arXiv:2607.Where the paper says it
Where the paper says it
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Introduction of bias-aware governance design: Implement mechanisms to make
choice order, author-choice visibility, and author-vote visibility should be documented and, where possible, made auditable rather than treated as neutral interface details.
This allows systems to explicitly expose the institutional design choices that influence voting power accumulation. -
Stance classification integration: Develop a stance classification procedure triangulated with external model and human evaluation checks to robustly classify choice stances into Approve, Reject, Abstain, or Other across various proposal formats. This ensures that
the qualitative findings remain the same
when comparing heuristic classifications with human evaluations. -
Bias decomposition for decision quality: Utilize regression models to separate conditional associations between voting-power share and position bias, approval bias, and author bias to understand
which channel is associated with a substantial, non-random concentration of support.
This allows for isolating whether high support is due to list ordering or author signals. -
Author influence transparency: Implement clear information disclosure requirements regarding
who created a proposal
andwhat option the author selected,
addressing concerns that large governance token holders can shape outcomes in ways thatordinary voters may not easily observe.
-
Interface design optimization: Adjust choice presentation based on empirical findings, such as ensuring that for proposals with two or three choices, an
Approve choice is almost placed in position 1
to maximize support capture.
Abstract
Decentralized Autonomous Organizations (DAOs) replace formal managerial authority with collective decision-making, by handing both proposal rights and token-weighted votes directly to community members. Yet, as in corporate votes, where management sets the agenda and its proposals rarely fail, DAO votes are strikingly one-sided. Using 127 281 proposals and about 51 million votes cast on Snapshot, the leading off-chain DAO voting platform, between 2020 and 2023, we ask which features of the ballot are associated with this support. We relate each choice's voting-power share to three observable features: its position in the list, whether its stance expresses approval, and whether the proposal author selected it. Each identifies one voting bias: position bias, approval bias, and author-selected choice bias. Author-selected choices show the strongest and most robust association, receiving a voting-power share 58.8 percentage points higher than non-author-selected ones, compared with 27.1 percentage points for approving choices and 7.7 percentage points for first-listed ones. About one third of the author bias comes from the author's own voting power; that advantage is smaller in the largest protocols and in proposals with higher participation. The results are robust across alternative specifications and outcomes. Our findings suggest that the agenda power that management holds in corporate voting persists even without a formal hierarchy, and that ballot design and the visibility of proposers' preferences are governance choices rather than neutral implementation details.
Sources
- Slaying the Dragon: The Quest for Democracy in Decentralized Autonomous Organizations (DAOs)
- Evaluating Position Bias in Large Language Model Recommendations
- A Study of Position Bias in Digital Library Recommender Systems
- The Vulnerable Nature of Decentralized Governance in DeFi
- The Hidden Shortcomings of (D)AOs -- An Empirical Study of On-Chain Governance
- Analyzing Voting Power in Decentralized Governance: Who controls DAOs?
- Disentangling the Effects of Social Signals
- How Decentralized is the Governance of Blockchain-based Finance: Empirical Evidence from four Governance Token Distributions
- Counterfactual Voting Adjustment for Quality Assessment and Fairer Voting in Online Platforms with Helpfulness Evaluation
- Understanding Blockchain Governance: Analyzing Decentralized Voting to Amend DeFi Smart Contracts
- Decentralized Finance, Centralized Ownership? An Iterative Mapping Process to Measure Protocol Token Distribution
- Who is Authoritative? Understanding Reputation Mechanisms in Quora
- Decentralization illusion in Decentralized Finance: Evidence from tokenized voting in MakerDAO polls
- An Empirical Study on Snapshot DAOs
- Strategic Vote Timing in Online Elections With Public Tallies
Related papers
- AI-Mediated Communication Can Steer Collective Opinion
- Epistemic Infrastructures of Science in AI Era Should Rebalance Costs of Generation and Verification
- AI out-persuades expert humans in text-based dialogue
- Evaluating Ambient Clinical Scribes in India: The Need for Multilingual Real-World Clinical Conversation Data
- AI Value Alignment for Evolving Social Norms
- GenAIT: Development and Validation of an Objective Generative AI Literacy Test for High School Students