When should groups make decisions through discussion alone, and when should they employ some kind of voting? This question has shaped how people govern themselves from the ancient world to modern corporations, and is now being debated in crypto. On one side, DAOs use token voting to make decisions. On the other, many layer-1 blockchain networks like Ethereum rely on rough consensus instead of voting. In the middle, projects like Solana use rough consensus along with a limited number of votes.
What do history and research tell us about whether to include explicit forms of voting, over and above deliberation and rough consensus? My answer: it depends.
Where groups can reach consensus quickly and trust one another, voting can create unnecessary vulnerabilities and encourage factionalism and obstructionism. But when groups become large or chaotic, voting provides an essential way to transparently measure the community’s preferences, bring legitimacy to the process, and close off long, unproductive debates.
Is voting on decisions useful? What’s come before and what’s going on in crypto today
Historical experiments with voting can inform us about whether — or perhaps when — voting is useful. In Athens and Rome, deliberation was important, but final decisions often required a vote to become official. Quaker Meetings, an important antecedent to America’s development of local democracy and a primary model for “town-hall democracy,” explicitly avoided voting and prioritized discussion until the group arrived at what they called the “Sense of the Meeting.”
More recently, and perhaps more directly relevant to crypto projects, open source software protocols have also diverged on the question of voting. The Apache Software Foundation and the Perl Steering Committee have both long employed formal voting after discussion to make important decisions. The Linux kernel development community, among others, prefers debate alone, with powerful figures — like Linus Torvalds, and other prominent founders or leaders —- making final decisions based on their sense of the group.
The Internet Engineering Task Force (IETF), an organization whose governance model directly influenced Ethereum’s, likewise avoids voting in favor of discussion. In a famous plenary session for the Task Force, computer scientist Dave Clark declared: “We reject: kings, presidents and voting. We believe in: rough consensus and running code.” This is also a mindset adopted by several others since, including the Ethereum community.
We can see the tension between deliberation and formal voting playing out today with major blockchain networks and DAOs. Consider three examples: Ethereum, Optimism, and Solana. While all three networks can trace their governance traditions to Bitcoin, Python, Linux, and broader open source projects, they diverge on this key question: Should formal voting supplement deliberative consensus?
At one end of the spectrum, Ethereum, like the IETF, rejects voting in favor of rough consensus. In the middle of the spectrum, Solana also leans on deliberation and rough consensus but includes a small number of explicit votes. These votes are purely advisory and occur only rarely, yet they can matter. At the other end of the spectrum, Optimism combines deliberation with formal, onchain voting to make binding decisions frequently, usually at least a few times a month and sometimes much more than that.
Rough consensus and voting: Better together … sometimes
Many software projects follow the old Quaker Meeting approach and use rough consensus with no voting. Despite how common this approach is in modern software development, remarkably little quantitatively oriented research evaluates how it works or when it might be preferred to voting systems. Indeed, most political economy graduate courses begin with models of voting, taking it for granted that this is the approach groups use to make decisions. Nevertheless, there might be some quite interesting advantages to using rough consensus without voting.
Bad arguments for rough consensus sans voting
Before we see the potential advantages of using rough consensus without voting, it’s helpful to first dispose of some popular but bad arguments in its favor, first. Proponents of rough consensus sometimes argue that through deliberation, participants may learn more about the issue and change their mind. While this is certainly true, it is not an argument against voting. Participants could deliberate, learn, and then vote at the end. So this is not really an argument for rough consensus versus voting as a means for making decisions; it is an argument for meaningful deliberation in either case.
Likewise, proponents often say that rough consensus is a way to encourage broader consensus in contrast to “majority rules” voting, which lets a small majority make decisions even if a large minority disagrees. But there is nothing to prevent a group from using voting with a rule other than majority rule — for example, the group could vote and require a unanimous vote to proceed. So this, too, is not really an argument in favor of rough consensus over voting.
Finally, sometimes people say that the ability to fork or exit the project obviates the need to vote. There are certainly cases where this is true, but much of the time, the cost of exiting is high, allowing many smaller disagreements to fester inefficiently.
Advantages to rough consensus without voting
Rough consensus incorporates preference intensity by letting participants scale up their vote through talking more and making stronger points in favor of their argument. In a formal voting system, everyone’s vote has a fixed weight (e.g., one person, one vote). With rough consensus and no voting, people who care more can have their views weighted more by leaning in and participating more in the conversation, and people who care less can opt out. While similar social processes could play out with voting, the formal weighting of each vote is a key constraint that is different from a purely deliberative process.
Similarly, rough consensus can allow for the informal weighting of some voices more than others. While a formal rule to weight some people’s votes more than others could lead to discord and gaming — in the spirit of Goodhart’s Law — rough consensus can give priority to more knowledgeable voices in an opaque way. By remaining opaque, it may be harder for people to game the system and gain outsize influence.
Rough consensus can create good will through the illusion of full agreement once the decision is made. By focusing only on discussion, rough consensus avoids a formal record of who supported or opposed the decision. This may help people who were more skeptical of the decision feel more bought into the decision even if they didn’t get their way, and since there is no record, they can even claim later to have supported the decision all along if it turns out well.
Rough consensus might avoid security vulnerabilities sometimes present in a permissionless, formal decision-making system. In the highly adversarial world of crypto, permissionless voting systems invite attackers to look for vulnerabilities. By requiring broad consensus with an opaque, human-driven process for determining when this consensus is reached, rough consensus may avoid some of these attack vectors.
Rough consensus can allow for dynamic decision thresholds. Instead of always requiring the same degree of agreement to make a decision, as in voting (e.g., pure majority, 2/3s majority, and so on), rough consensus allows for the group to require higher levels of debate and consensus on more important or more contentious topics. This can be more efficient, and it can also create uncertainty that could be helpful for giving people incentives to continue leaning in even if they would be confident of succeeding at a static decision threshold. Similarly, it also does not force a binary decision in a finite period of time the way formal voting does. Instead, the discussion can continue and the decision can remain in suspended animation, allowing a minority with strong views to delay the decision and, in another way, effectively increasing the threshold to make the decision.
Rough consensus avoids formally declaring who is a stakeholder. By inviting a broad range of interested actors to participate in the discussion, and by avoiding a formal decision aggregation rule, rough consensus does not have to spell out explicitly who has how much of a say in the decision. This helps avoid accidentally excluding valuable stakeholders, and also might encourage political harmony.
Advantages to adding voting on top of rough consensus
Despite the many potential benefits of rough consensus outlined above, most communities across the world adopt formal voting procedures at some point. To be fair, most of these communities are not organized around software development. Nevertheless, as communities grow in scale and complexity, voting becomes increasingly appealing.
Voting can provide a more accurate assessment of the community’s views. If it works well, voting provides a credible signal of the whole community’s views that doesn’t weight decisions towards the loudest voices. Deliberation, on the other hand, might give priority to those willing to talk more, even if they have unrepresentative or uninformed views. This becomes especially important as the community becomes large, so that dialogue becomes increasingly unreliable as a way to gauge mass sentiment.
For this to hold, however, participation in voting must be broad. As we know, participation in token voting today does not always meet this threshold (though participation in rough consensus isn’t clearly superior).
Voting keeps decisions moving. Formal voting is also a way to close off debate and make a final decision that the group understands won’t be revisited. Without strong norms, pure deliberation can go on for a much longer time. As Apache Foundation documents note, people “tend to avoid conflict and thrash around looking for something to substitute — somebody in charge, a rule, a process, stagnation. None of these tend to be very good substitutes for doing the hard work of resolving the conflict.” Similarly, Vitalik Buterin has acknowledged that “Ethereum governance has the problem that it does not have a mechanism for conclusively rejecting things.” Voting forces the hard work to reach resolution, and helps conclude debate on rejected issues.
Voting lends transparency to the decision-making process. A formal vote, if conducted through a transparent process such as onchain voting, removes some doubt about whether the decision was made fairly or not. In contrast, in rough consensus systems, a small group of powerful actors may be in a privileged position to assert that consensus has been reached even if it hasn’t. Likewise, while avoiding declaring which stakeholders have how much say might be an advantage to rough consensus, it may also let gatekeepers encourage a fiction, claiming that more stakeholders have power than is actually true. Voting helps to dispel this illusion.
Voting, though, is not a magical solution. Votes, too, can serve as governance theater. If participation is uneven, or if the vote is dominated by a small group of powerful delegates whose views do not align with the broader community, a vote can appear to reflect more voices than it actually does. In this way, both voting and rough consensus face a similar legitimacy problem.
How much voting?
As I mentioned above, the Solana validator community carries out a small number of votes focused on key decisions, and these votes are merely advisory. Moreover, it is not entirely clear when a decision needs to be referred for a vote, and there have been recent debates centered around this challenge. By signaling publicly the views of validators (who serve as the voters in Solana’s governance system), these votes, when they do occur, can and have influenced outcomes, however.
Many DAOs, like Optimism, use voting more frequently, requiring tokenholders to vote on important economic decisions like treasury management and business operations, and giving them veto rights over other decisions like protocol upgrades. In this more aggressive system of voting, what decisions require votes is more explicit, and the votes are not just advisory but instead binding.
If a project determines that it cannot rely on rough consensus and needs to use voting, how much voting should it use? Should it adopt a system more like Solana’s, or more like Optimism’s? Again, it depends.
The more confident a project is that there are trusted contributors qualified to assess the rough consensus of the community and to decide when advisory votes are necessary, the less valuable voting will be. Because voting brings with it vulnerabilities and procedural delays, minimizing it can be useful.
On the other hand, when communities cannot or do not want to trust leaders to choose when to refer decisions to votes or to honor advisory votes, formal voting becomes increasingly valuable.
Finally, it may be that the layer at which the project operates may matter. As a Layer-2 blockchain network, Optimism is an object within the Ethereum system. This might make it harder for users to exit if they disagree with a decision, which in turn might mean that formal voting becomes especially important. On the other hand, it is a conspicuous pattern that almost all of the major Layer-1 blockchains do not use formal voting to the same extent that DAOs do.
The role of culture
How well deliberation, rough consensus, and formal voting work also depends heavily on underlying cultural conditions. Deliberation-only systems require sustained norms of good-faith engagement, reasonable timelines for reaching consensus, and community trust in process leaders who drive the proposal process and get to assert when rough consensus has been reached. When these conditions hold, formal voting may introduce unnecessary friction, encourage divisiveness, and open up potential attack vectors.
Cultural foundations also shape how core principles and constitutions matter. Healthy communities can use appeals to values as a powerful check on overreach, effectively raising the bar for decisions that would violate founding commitments. In less healthy ones, however, small groups with disproportionate influence may invoke principles selectively in order to block decisions they oppose, raising the deeper question of how consensus arises and who has the informal authority to declare it.
But cultural foundations can shift. As networks grow, stakeholders become more diverse, and technical complexity deepens, the informal mechanisms that enable rapid deliberative consensus may strain or break down. The influence of a small number of major voices may appear less legitimate to a broader community. Consensus timelines may stretch as more voices demand to be heard.
Under these conditions, formal voting provides the capacity to make decisions even when cultural consensus mechanisms falter. The question for any blockchain governance system is not whether voting is inherently better than deliberation, but rather: What institutional arrangements will be most effective for the particular realities of the network and its community?
***
Ethereum, Solana, Optimism, and many other projects are conducting live experiments in this fundamental governance challenge. The results will offer helpful lessons not just for blockchain networks, but for any decentralized system grappling with the tension between deliberative ideals and institutional realities.
***
Acknowledgment: Many thanks to Peter Buisseret, Vitalik Buterin, Scott Kominers, Justine Lavande, Eliza Oak, Tim Roughgarden, and Tomasz Stanczak for comments, and especially to Pranav Garimidi for suggesting this line of research.
***
Andrew Hall is the Davies Family Professor of Political Economy in the Graduate School of Business at Stanford University and a Senior Fellow at the Hoover Institution. He works with the a16z research lab and is an advisor to tech companies, startups, and blockchain protocols on issues at the intersection of technology, governance, and society.
***
The views expressed here are those of the individual AH Capital Management, L.L.C. (“a16z”) personnel quoted and are not the views of a16z or its affiliates. Certain information contained in here has been obtained from third-party sources, including from portfolio companies of funds managed by a16z. While taken from sources believed to be reliable, a16z has not independently verified such information and makes no representations about the current or enduring accuracy of the information or its appropriateness for a given situation. In addition, this content may include third-party advertisements; a16z has not reviewed such advertisements and does not endorse any advertising content contained therein.
The views expressed here are those of the individual AH Capital Management, L.L.C. (“a16z”) personnel quoted and are not the views of a16z or its affiliates. Certain information contained in here has been obtained from third-party sources, including from portfolio companies of funds managed by a16z. While taken from sources believed to be reliable, a16z has not independently verified such information and makes no representations about the current or enduring accuracy of the information or its appropriateness for a given situation. In addition, this content may include third-party advertisements; a16z has not reviewed such advertisements and does not endorse any advertising content contained therein.
You should consult your own advisers as to those matters. References to any securities or digital assets are for illustrative purposes only, and do not constitute an investment recommendation or offer to provide investment advisory services. Furthermore, this content is not directed at nor intended for use by any investors or prospective investors, and may not under any circumstances be relied upon when making a decision to invest in any fund managed by a16z. (An offering to invest in an a16z fund will be made only by the private placement memorandum, subscription agreement, and other relevant documentation of any such fund and should be read in their entirety.) Any investments or portfolio companies mentioned, referred to, or described are not representative of all investments in vehicles managed by a16z, and there can be no assurance that the investments will be profitable or that other investments made in the future will have similar characteristics or results. A list of investments made by funds managed by Andreessen Horowitz (excluding investments for which the issuer has not provided permission for a16z to disclose publicly as well as unannounced investments in publicly traded digital assets) is available at https://a16z.com/investment-list/.
The content speaks only as of the date indicated. Any projections, estimates, forecasts, targets, prospects, and/or opinions expressed in these materials are subject to change without notice and may differ or be contrary to opinions expressed by others. Please see https://a16z.com/disclosures/ for additional important information.
