My name is Andy Hall, I’m a political economy professor at Stanford where my lab studies how to build effective systems of democratic governance. I’m also an advisor in tech where I work on solutions to online governance problems. I’ve become interested in blockchain governance and “DeGov,” and have started to study these topics academically. Through this research I have been drawn to Maker as a particularly successful example of DAO governance.
Over the past several months, I have been lurking a bit on this forum, chatting with some of you on Discord, and working with @0xdeniz to study Maker governance.
Goal: Explore Research-Driven Insights for Maker Governance
Our goal is to see if insights from the long-running study of “physical-world” governance are relevant for DAO governance and can help Maker as it thinks about new governance solutions in the context of the clean money vision, the Sagittarius engine, and the natural evolution of the DAO.
Having done some early thinking, we thought it was a good time to bring the project to the community and get thoughts from people who might be interested before developing potential next steps.
After looking at governance from many angles, we have homed in on three central challenges to effective decentralized governance:
- Aligning incentives. Aligning the incentives of the Maker workforce with MKR holders, as @rune discussed in depth here.
- Participation. Increasing MKR holder participation in governance, as @0xdeniz discussed here.
- Information and expertise. Making smart, informed decisions about complex, technical issues in a context where most MKR holders are unlikely to have the necessary time or expertise. (This issue is raised in many forum discussions including the discussion linked above in item 2.)
These three challenges are core to democratic governance, and there are insights from research about all three that may be helpful to Maker and to the general design of DAO governance.
The future success of Maker, as it grows and evolves, depends on continuing to align the incentives of the workforce with the preferences of the MKR holders, a challenge common to many types of organizations. Incentive alignment is not only important for getting workforces to focus their efforts on maximizing value for shareholders; it is also important for investment, as investors will presumably prefer to commit resources to projects they believe are sufficiently well governed to not waste or steal their resources. This is why good governance is important for growth in a broad variety of contexts and is a classic topic in political economy.
In democratic systems, this is all about getting politicians to do a good job, and in political economy we often boil this down into two parts:
- Choosing honest, competent delegates;
- Giving delegates incentives to work hard for their constituents.
This idea is well expressed in James Madison’s Federalist 57, where he writes: “The aim of every political constitution is, or ought to be, first to obtain for rulers men who possess most wisdom to discern, and most virtue to pursue, the common good of the society; and in the next place, to take the most effectual precautions for keeping them virtuous whilst they continue to hold their public trust.”
For Maker, there are at least two layers of principal-agent relationships to think about: the relationship between the workforce and MKR holders, and the relationship between MKR holders and their delegates, who can help them to monitor the workforce. This is analogous to standard legislative and bureaucratic politics in a democracy: voters elect representatives (like delegates), and those representatives are responsible for monitoring and overseeing the bureaucracy (like the workforce).
How do political systems select good delegates and give them incentives to properly oversee the workforce? Solutions from democratic governance that may be relevant to Maker (and other DAOs) include:
- Compensation of delegates. As has been discussed recently on the forum, it is important to build good compensation structures for delegates that encourage good people to become delegates, and give them incentives to work hard while in office and to desire reelection. [For example, I wrote a book about why paying politicians too little, among other factors, can contribute to polarization by deterring moderate, sensible people from becoming politicians. For some further key tradeoffs involved, see this famous paper by Tim Besley. For arguments against paying politicians, take a look at this underappreciated essay by Benjamin Franklin.]
- Periodic elections. If being a delegate is sufficiently valuable to people, then periodic elections can help make delegates think forward to future reelection bids, causing them to work hard now because they want to win reelection later. [For example, I recently published a paper about how having to face reelection makes US state legislators show up for more votes, serve on more committees, and write more bills.]
- Legislative oversight. Designing a system for the delegates that allows them to divide labor among themselves, develop specialized expertise on particular issues facing the DAO, and monitor different parts of the workforce. This could include full-time staff for delegates, a committee system, etc. [For a good overview of academic thinking on legislative organization, see this review paper by Keith Krehbiel.]
Properly empowered delegates can in turn align the incentives of the workforce with the preferences of MKR holders by developing policies related to how the workforce is compensated and monitored. But this only works well with a sufficiently well designed delegate system.
Increasing MKR-holder participation—in terms of discussing, voting, delegating, and serving as a delegate or other contributor—seems important for a number of reasons. Insufficient participation could lead to poor monitoring of the workforce and misaligned incentives. Just the possibility of concerns like these, whether true in practice or not, could scare away potential investors or users who worry about their money being misused. It may also leave the system more vulnerable to governance attacks, among other issues, as research about Maker by some of my Stanford colleagues explores.
Getting people to participate is always difficult, because of the fundamental free-rider problem: as a token holder, I enjoy the benefits of the Maker system whether or not I participate. Since participating takes time and effort, and since my participation generally has little or no impact on the outcome, I have incentives not to participate. This is particularly true for small token holders, given the nature of token voting. Perhaps for this reason, we see low rates of participation in Maker governance, as we do in many physical-world settings (examples include low rates of voter participation in many political elections; the difficulty of getting people to run for office in many democracies; and the low rates of participation in shareholder governance for publicly traded corporations.)
On the other hand, many democratic systems overcome this challenge, and surprisingly large numbers of people participate in many voting systems even though it is difficult, if not impossible, to articulate rational self-interested reasons for them to do so. This is often referred to as the “paradox of voting.” We see similar patterns in other forms of online participation, like Wikipedia or Reddit, where skeptics used to think—clearly incorrectly—that it would be impossible to get communities to participate in maintaining the community.
Experiences in democratic elections suggest three important insights for Maker as it considers ways to expand participation:
- Reducing costs of voting is important, but not enough. Very high costs of voting clearly dissuade participation (see for example this research on how the southern US raised voting costs to suppress black turnout), but lowering costs without making other changes often fails to increase participation meaningfully. For example, my lab recently published a study of the 2020 US election and the largescale shift to mail voting, which many people thought would increase turnout dramatically by making it easier to vote from home. It had almost no effect on turnout, because another fundamental barrier to voting in US elections has to do with feelings of disempowerment and disinterest with the system, which is not affected by lowering the costs of voting.
- People tend to participate more when they care, intrinsically or socially. Caring about the issues at hand, being part of a network of people who are paying attention to the issues, and social pressures are important drivers of participation, beyond the costs. [One good example of this is a famous experiment showing that threatening to reveal your non-voting to your neighbors increased turnout dramatically in a US primary election.]
- People tend to vote more when they understand the decision they’re facing. Lowering the cognitive costs of voting is also important. We think, for example, that increasing education leads people to vote more. Simpler voting systems that reduce the complexity of the number and difficulty of voting decisions are likely to increase participation, too.
Decentralized governance works best when the governing community understands the issues it is debating and voting on. This is perhaps the oldest concern about democracy, stemming back to Plato, Aristotle, and many other critics of democracy in Ancient Greece.
Experience in a wide variety of democratic systems suggests that there are at least four core elements that help imperfectly mitigate this problem while preserving some of the benefits of elections and voting:
- Simplify the voting problem. If you look at how most polities run their elections, or how shareholder voting works for publicly traded companies, votes typically occur at a set, relatively infrequent interval, and concern a relatively small number of voting decisions. While limiting the number of things that are voted on may seem like it reduces how “democratic” a system is, it can actually increase it if it empowers voters by giving them a clearer and simpler set of decisions to make. A great book to read along these lines is Riker’s Liberalism Against Populism, which argues that the point of elections is simply to let voters control the hiring and firing of key political officials—not to understand and weigh in on each specific decision that government makes.
- Vote on delegates more than on decisions. Most democratic systems rely on asking voters to choose a relatively small number of key representatives. Those delegates take charge of identifying and appointing experts. This creates one layer of insulation between experts and elections. We tend to think this is helpful for at least two reasons: first, it helps us find and retain experts who may not be interested in, or have the particular skills for, winning elections; and second, it shelters the experts at least partially from direct electoral pressures that might lead them to pander and take bad but popular decisions. [For some key tradeoffs, see this essay on elected vs. appointed judges].
- Rely on media to help inform voters. An ecosystem of experts who monitor the system and can alert voters when something concerns them (or when something is going particularly well) can help create accountability. The news media is thought to be especially important in performing this duty in physical-world governance [see for example this great paper showing how congressional districts in the US that for somewhat random reasons get more news coverage about politics end up knowing more about politics, paying more attention to their politicians, and ultimately have members of Congress who work harder for them]; it is not clear yet how this void will be filled in DeGov, but it seems like an important opportunity.
- Structure political conflict with parties and interest groups. Parties and interest groups distill complex issues into simple decisions for voters and create a potential layer of accountability. Similar to how the media can inform voters, when parties or interest groups support a particular delegate or a vote on a particular issue, it can signal to voters who share their preferences that this is something they should support. In this way, it can simplify the problem for voters, who can pay the relatively low cost of figuring out which parties or interest groups are aligned with them, and then try to trust their endorsements concerning some set of future decisions. [The classic case for parties and ideology as informational short-cuts for voters comes from Anthony Downs’s An Economic Theory of Democracy.] This is obviously an imperfect system and is difficult to create from scratch, but we know that most election-based societies evolve over time to have quite well developed party and interest group systems.
The ideas above hopefully give some sense of how insights from political economy might apply to DAO governance and to Maker, specifically. We are curious for people’s thoughts on these ideas, whether they are valuable, and where we might take this project next. One potential next step could be to develop specific design recommendations for Maker based on these ideas and share them in a next memo.
For my own part, I plan to pursue academic research around DAO governance in general, but am also excited to talk with people about Maker specifically and to explore if there are constructive ways for me to help build new forms of governance. Please do not hesitate to reach out to me if you’d like to chat.