Introducing KeeperDAO Governance

Introducing KeeperDAO Governance

The purpose of this document is to lay the foundations for the community governance of KeeperDAO. It draws inspiration from a vast body of work, reaching beyond the traditional blockchain literature in order to give the governors of KeeperDAO the highest standard of knowledge about the new and extraordinary system they will be responsible for.


What is KeeperDAO?

KeeperDAO seeks to attract independent Keepers and coordinate their access to certain shared resources, namely various blockchain networks and settlement layers such as Ethereum. This coordination ensures that when Keepers use the shared resources to generate profit, for example by executing an arbitrage or liquidation, the profit is captured efficiently, and not destroyed by contention or the extraction of MEV.

In addition to this, KeeperDAO designs and builds its own unique products for use by individual and institutional users, and routes the orders from these products through its community of Keepers for their exclusive execution.

We call these two strategies the Coordination Game, and the Hiding Game, respectively [75]. The first increases Keepers' expected profitability by eliminating costly competition, and the second gives participating Keepers more opportunities to profit.

Figure 1: A schematic overview of the KeeperDAO ecosystem.

Why is Governance Important for Us?

The KeeperDAO ecosystem is voluntary, but its design is rooted in the theory of coordination games and economic self-interest, meaning we can still predict outcomes. The theory tells us how important it is to enforce certain "golden rules," in order for the system to be self-sustaining. Among these rules are:

  • Participation in KeeperDAO increases a Keeper's expected payoff.
  • Participating Keepers can't get away with not coordinating.
  • Participating Keepers can't get away with manipulating user orders.

To ensure these rules hold, we must constantly monitor and adapt to new technology and products, new forms of MEV, and an ever-changing cast of users, Keepers, and markets, while continuing to expand and develop our own unique technology.

This effort, in a nutshell, is what the governors of KeeperDAO, equipped with their governance process and all its technology and procedures, will achieve. Governance is the engine that will take in our collective knowledge and ideas, use them to upgrade our protocol and products, and regulate the Keepers executing the transactions. Our success rests on our ability to make this engine run efficiently. If we can do this, our protocol will thrive.

Getting to Governance

Design Principles

Systems of governance should be tailored to fit their communities and the problems they are intended to solve. Rather than select what is fashionable, or mimic the governance of others, it is our responsibility to create the governance that, above all, is best for KeeperDAO.

Instead of buzzwords, we start from known facts, and follow a process of deduction to derive a promising form of governance that meets our needs and is supported by decades, sometimes millennia, of scholarly research. You, reader, now form the newest link in this chain of knowledge. Directly or indirectly, you are now a governor of a collective institution, and have a role to play in shaping its form. Together, we will cultivate a new and highly productive system of decentralised infrastructure — the Keeper economy.

The Keeper Economy

The Keeper economy consists of independent, automated agents called Keepers who control their own liquidity on the blockchain. Keepers use their store of assets to execute transactions in exchange for profitable incentives, like the price imbalances of arbitrage, or the discounted collateral offered for liquidation.

Sometimes portrayed as opportunistic “bounty hunters,” only interested in guaranteed profit, Keepers are in fact an essential part of a functioning decentralised economy [59,60,61,62]. Their ability to analyze and transact across networks leads to more efficient prices and rates, and by acting as a buyer of last resort, they remove risk from decentralised lending protocols and can backstop the health of the entire system in a crisis. They excel at all forms of automation requiring liquidity, and their ability to execute sophisticated transactions in adverse conditions makes them ideal vehicles for many kinds of complex services.

Today, the Keeper economy is suffering due to the effects of miner extractable value, or MEV [63,64]. The erosion of their profitability due to Priority Gas Auctions and related factors means that their share of the pie has decreased rapidly, even as the need for their services has skyrocketed. Worse, the rise of predatory front-running and sandwich attacks has left users justifiably wary of their presence on the network, leading to its infamous description as a “Dark Forest” [65].

Common-Pool Resources and Ostrom's Proto-DAO

The plight of Keepers and DeFi users is not so new. It bears strong resemblance to classical economic problems like the prevention of overfishing. Driven by their own self-interest, individual fishermen are strongly incentivised to overexploit the common resource of the fishery, and end up extinguishing their own livelihoods in a race to the bottom.

For nearly 200 years, it was accepted that only top-down state intervention could prevent such "tragedies of the commons," [66,67] until economist Elinor Ostrom demonstrated, in landmark studies of local fisheries and similar systems around the world, that ordinary people are capable of creating rules and institutions that allow for the sustainable and equitable management of common-pool resources [6]. For this work she was awarded the 2009 Nobel Prize in Economics, only 19 days before Nakamoto's publication of the Bitcoin whitepaper [68].

Ostrom and her colleaguees argue that there is a significant class of problems, including the management of collective or public goods, where it is better to rely on more "decentralised" forms of management, in an approach "based on the assumption that we need to concentrate on voluntary actions and the processes and institutions that emerge from them", and "presuming a certain skepticism about the power of coercion" [69,70,55,56].

The type of institution that Ostrom describes is what we might today call a Decentralised Autonomous Organization, or DAO [71]. In other words, to the extent that Keepers face a problem of the commons, the evidence suggests that a Keeper DAO could fix it.

Consensus Democracy

It would be a mistake to assume that DAOs, or the institutions described by Ostrom, are simply loose federations held together by goodwill, norms, and shared values, or that the power of their law comes from (smart) contracts alone. There are deeper forces that bind collectives together.

A DAO is a group whose membership is determined by ownership of the DAO's governance token on a permissionless blockchain. This means that every member of the DAO has at least one right: the right to transfer their governance tokens [72].

Already, this gives members the power to grant membership (by transferring their tokens to a non-member), and to revoke their own membership (by transferring all of their tokens to someone else). Revoking your own membership, i.e. leaving the group, is an important concept called exit [3,22,45,47].

DAOs are bound together by the power of exit, because for DAOs, exit has consequences. Their open source software, transparent processes, and permissionless nature means that they live under the constant "shadow of the fork": if enough members exit, they could fork the software and replicate the DAO with their own ideas, instead of the ones they left behind.

This splitting of the pie is usually in nobody's interest, so when there is a significant threat to exit, "remainers" will act to prevent the exit from happening. But what can they do? Voting won't work. The exiteer is autonomous; they will simply leave if they don't like the outcome of the vote. In the words of John Locke, the remainers are "bereft of compulsive power", and can only convince the exiteer to remain using  "exhortations, admonitions, and advices" [49].

This kind of governance, using "words, not votes," characterizes what is called consensus democracy, a decision-making structure which involves and takes into account as broad a range of opinions as possible, as opposed to systems where minority opinions can potentially be ignored by vote-winning majorities.

Urfalini's Rule of Non-Opposition

How is consensus democracy practiced? Surprisingly, it follows a fairly consistent pattern, even though it arises spontaneously in many different contexts around the world.

Barbara Yngvesson [19] spent a year observing the decision-making processes in a community of fishermen on a small island off the west coast of Sweden. She found that when making an important decision during an expedition, such as changing the fishing location, the process was three-fold:

  1. One of the fisherman suggests moving to another location.
  2. During the following half-hour or so, the others may express reactions to this proposal.
  3. In the case of no counter-proposal, the first speaker reiterates his proposal and the boat heads for the designated fishing area.

Philippe Urfalino calls this the "Rule of Non-Opposition", and contrasts it with voting as a method of consensus decision-making [13,10]. It is also sometimes called a "silence procedure", or described with the Latin phrase qui tacet consentire videtur "who is silent is seen to agree".

The rule of non-opposition appears everywhere in deliberative situations across the globe, from Sudanese tribal councils to Swiss cantons to the World Trade Organisation [53,54,73]. One especially relevant practitioner is the Internet Engineering Task Force (IETF), which for decades has successfully used the method of "rough consensus" for group decisionmaking [50,51].

"Our credo is that we don't let a single individual dictate decisions (a king or president), nor should decisions be made by a vote, nor do we want decisions to be made in a vacuum without practical experience. Instead, we strive to make our decisions by the consent of all participants, though allowing for some dissent (rough consensus), and to have the actual products of engineering (running code) trump theoretical designs." (IETF RFC7282)

Rough consensus, along with the "Request for Comment" (RFC) proposal format, are IETF innovations and the ancestor of much of the protocol governance practiced today. The widely-imitated Ethereum Improvement Proposal (EIP), for instance, was based on the Bitcoin Improvement Proposal (BIP), which in turn was based on the Python Enhancement Proposal (PEP), which takes significant inspiration from the IETF RFC. And similar ideas of "rough consensus" are a major part of what powers the soft governance philosophy of Ethereum [4].

Rethinking the Token-Weighted Vote

So then, when and why did voting get involved in DAO governance? It was likely borrowed in by "seductive analogy" with corporate shareholder voting [46]. In corporate governance, though, the outcome of a shareholder vote is enforceable by federal laws guaranteeing the rights of corporate shareholders. No such laws, and hence no such rights, exist for members of a DAO. Ostrom has shown that such "anchoring" to state law is neither necessary nor always desirable, a view in keeping with the philosophy of decentralisation.

If the outcome is not enforceable, does it make sense to consider a tokenholder vote as having "passed" or "failed"? Probably not. And at a deeper level, doesn't the right of exit mean that a vote has no way to prevent the negative externalities of the DAO losing consensus? The corporate entities most similar to a DAO in this regard are mutual funds and SPACs, and in these entities voting has become vestigal or even been eliminated [44,45,46,47], a fact mirrored in the MolochDAO design [52]. We do not even need to dip into questioning oligarchic or plutocratic forms, or the governance attacks they engender, to see that most tokenholder voting is not analagous to shareholder voting at all.

Instead, we suggest that tokenholder voting, or more precisely token-weighted voting, measures the scale of a threat to exit. When considering a proposal, those who would exit if the proposal were to pass, may signal that fact using a token-weighted vote that only counts "no"s. The result is the number of tokens that would exit, allowing the group to measure the degree of consensus. In other words, we propose that the natural use for a token-weighted vote is to answer the question, "are there any objections?".

Coordinators, Experts, and Meritocracy

Even with an egalitarian form of consensus governance, KeeperDAO is embedded in, and interacting with, a highly competitive market environment. It is a "production community," where members collaborate in the creation or maintenance of collective goods, and "own" the output of their labor [33].

Scholarship on open source production communities suggests that governance must be meritocratic in order to attract high quality contributions from voluntary members [32,37,74]. But in meritocratic institutions, power can be earned through reputation or contribution, which can lead to the establishment of distinguished individuals. Does the presence of such "distinguished individuals" harm the legitimacy of decentralised governance? The evidence suggests the answer is no, as long as their authority is properly limited, and that in fact they can play a vital role.

Production communities need a way to manage their interdependence in order to direct individual efforts toward a common goal [32,28]. This need leads to the emergence of limited positional authority for some individuals, allowing them to facilitate coordination and decisions, aid in conflict resolution, and represent the project to outside parties. While most authority remains laterally distributed, truly "flat" structures are very rare [57,34,35,36].

This limited positional authority is part of what Buterin calls the "coordination institution" [4], and it appears also in fisheries, where Gutiérrez, Hilborn and Defeo found that legitimate community leaders, when guided by collective interests and not self benefits, give resilience to changes in governance, and enhance conflict resolutions, concluding that the presence of community leadership was critical for successful co-management [54].


  • The Keeper economy is suffering from a problem of the commons.
  • Ostrom's work shows that such problems have their solution in decentralised co-management and co-ownership of the public good.
  • A DAO embodies most of Ostrom's principles for such an institution, in particular it has an institutionalised exit.
  • Institutionalized exit renders the compulsive power of voting meaningless, and instead supports a consensus method of governance.
  • The most widely-practiced form of consensus governance, from Swedish fishing boats to the IETF, is what Urfalino calls the "rule of non-opposition".
  • On a meritocratic basis, limited positional authority for coordination makes production communities more competitive.

KeeperDAO governance framework

General attributes

All governance models involve trade-offs between a set of strengths and a set of weaknesses. Balancing different governance modes against each-other can allow them to cancel out weaknesses while preserving strengths, at the cost of added complexity. Complexity is the enemy — it drives down efficiency and decreases participation (hence legitimacy).

We introduce a minimalistic three-mode governance designed so that when consensus is high, efficiency is high, and when consensus is low, efficiency is low. This allows our governance to be as efficient as possible, while maintaining a stable level of legitimacy. Not only does it mean that we take our time on hard problems so nothing is "rammed through", but it creates an additional incentive — when we preserve a high degree of consensus, we are more efficient, and hence more competitive.

Figure 2: (Left to right) Non-adaptive governance is inefficient when consensus is high; efficiency of adaptive governance changes with level of consensus; relationship of individuals involved with the DAO in some way.


The formation of new proposals is the most important process in the DAO, and we believe anybody should be able to draft or comment on them. By making a clear, simple process that is open to all, we increase the likelihood that beneficial ideas will find their way to us in the best form.

Proposals are structured in a standard format called a Keeper Improvement Proposal, or KIP. Draft proposals are posted by their authors on the open governance forum, and are eventually merged into the permanent governance repository.

Once a proposal is drafted, the DAO will attempt to reach consensus on the proposal. If consensus is reached, the proposal becomes part of the current state of the DAO.

Consensus Procedure

Consensus takes place in three stages, each involving a different method. The status of a proposal indicates what stage of the proposal process it is in.

Figure 3: Proposal statuses, transitions, and consensus stages.

Stage 1 - Crowd Consensus

After a draft proposal is posted on the governance forum, crowd consensus takes place. Members of the governance forum or general public may comment on, ask questions, and work with the author to revise or refine the proposal. This avoids overloading the governance system with proposals, and draws on the wisdom of the crowd and the scalability of its attention.

When the proposal author judges that consensus has been reached, they may move the draft proposal into the next phase of the consensus, by opening a pull request into the governance repository.

Stage 2 - Expert Consensus

During expert consensus, members of the DAO with the designation proposal editor merge the pull request for the proposal, then conduct an independent review of its contents. The output of this review is a joint recommendation for the DAO to adopt, decline, or defer the proposal, along with the reasoning behind that conclusion.

Publication of this joint recommendation moves the draft proposal to the next and final phase of the consensus.

Stage 3 - Tokenholder Consensus

During tokenholder consensus, members of the DAO may signal objection to the recommendation of the editors. This objection is recorded as a simple token-weighted vote which only counts "no"s. There is no quorum, and no decision function, only a magnitude.

Members of the DAO who object must make their own recommendation and objection known, along with what, if anything, could be altered in the proposal in order to gain their approval.

Challenges and Open Questions

Although this type of governance is not new, it blends existing knowledge and practices in a new way. We cannot know whether it will be perfectly effective, and there are some questions that we have left open, to allow us to debate, discuss, and collectively shape our preferred solution.

What is certain is that the challenges MEV presents to users, Keepers, and the decentralised financial system as a whole, are not going away. We can wait for state regulation, pin hopes on a technological arms race, or we can create a protocol that allows individuals to act in their own self-interest and still generate a public good. A walled garden to overgrow the dark forest, a meeting-place open to all good Keepers, builders, partners, and users, owned by the community and managed as a decentralised resource for all.

That is KeeperDAO.


  1. Freeman, J. "The Tyranny of Structurelessness" (1970)
  2. Coase, R. H. "The Nature of the Firm", Economica (1937)
  3. Hirschman, A. O. "Exit, Voice, and Loyalty: Responses to Decline in Firms, Organizations, and States" (1970)
  4. Buterin, V. "Notes on Blockchain Governance" (2017)
  5. Buterin, V. "On Collusion" (2019)
  6. Ostrom, E. "Governing the Commons: The Evolution of Institutions for Collective Action" (2015)
  7. Alston, E. and Law, W. and Murtazashvili, I. and Weiss, M. B. H. "Can Permissionless Blockchains Avoid Governance and the Law?" (2020)
  8. Lee, B. E. and Moroz, D. J. and Parkes, D. C. "The Political Economy of Blockchain Governance" (2020)
  9. Stifter, N. and Judmayer, A. and Schindler, P. and Kern, A. and Fdhila, W. "What is Meant by Permissionless Blockchains?" (2021)
  10. Urfalino, P. "Apparent Consensus and Vote: Two Modes of Collective Decision-Making" (2006)
  11. Urfalino, P. "La réalité des groupes agents" ["The Reality of Group Agents"] (2017)
  12. List, C. and Pettit, P. "Group Agency: The Possibility, Design, and Status of Corporate Agents" (2011)
  13. Urfalino, P. "The Rule of Non-Opposition: Opening Up Decision-Making by Consensus" (2014)
  14. "Wikipedia:Consensus"
  15. Babenko, I. and Choi, G. and Sen, R. "Management (of) Proposals" (2019)
  16. Yngvesson, B. "Leadership and consensus: Decision-making in an egalitarian community" (1978)
  17. Charléty, P. and Fagart, M. and Souam, S. "Quorum rules and shareholder voting" (2019)
  18. Aguiar-Conraria, L. and Magalhães, P. C. and Vanberg, C. A. "Experimental evidence that quorum rules discourage turnout and promote election boycotts" (2016)
  19. Aguiar-Conraria, L. and Magalhães, P. C. and Vanberg, C. A. "What are the best quorum rules? A laboratory investigation" (2020)
  20. Aguiar-Conraria, L. and Magalhães, P. C. "Referendum design, quorum rules and turnout" (2010)
  21. Herrera, H. and Mattozi, A. "Quorum and Turnout in Referenda" (2010)
  22. Newman, D. G. "Exit, Voice, and 'Exile': Rights to Exit and Rights to Eject" (2007)
  23. Clark, W. R. and Golder, M. and Golder, S. N. "Power and Politics: Insights from an Exit, Voice, and Loyalty Game" (20??)
  24. Clark, W. R. and Golder, M. and Golder, S. N. "An Exit, Voice and Loyalty Model of Politics" (2017)
  25. Camerer, C. F. and Weigelt, K. "Experimental Tests of a Sequential Equilibrium Reputation Model" (1988)
  26. Schulz, E. and Konstantinidis, E. and Speekenbrink, M. "Learning and decisions in contextual multi-armed bandit tasks" (2015)
  27. Reverdy, P. and Srivastava, V. and Leonard, N. E. "Modeling Human Decision-making in Generalized Gaussian Multi-armed Bandits" (2019)
  28. Magrassi, P. "Free and Open-Source Software is not an Emerging Property but Rather the Result of Studied Design" (2010)
  29. Harris, L. "The Politics of Shareholder Voting" (2011)
  30. Rea, A. and Kronovet, D. and Fischer, A. and du Rose, J. "COLONY Technical White Paper" (2020)
  31. Merkle, R. C. "DAOs, Democracy, and Governance" (2021)
  32. O'Mahony, S. and Ferraro, F. "The Emergence of Governance in an Open Source Community" (2007)
  33. O'Neil, M. and Pentzold, C. and Toupin, S. "The Handbook of Peer Production" (2021)
  34. Moffett, M.W. and Garnier, S. and Eisenhardt, K.M. et al. "Ant colonies: building complex organizations with minuscule brains and no leaders" (2021)
  35. Marsh, H. "Stigmergy"
  36. Waters-Lynch, J. and Duff, C. "Coworking's Cooperation Paradox: On the Role of Stigmergic Curation" (2021)
  37. Hall, D. M. and Bonanomi, M. "Governing Collaborative Project Delivery as a Common-Pool Resource Scenario" (2021)
  38. "DAOstack An Operating System for Collective Intelligence White Paper V1.1" (2018)
  39. Hanson, R. "Shall We Vote on Values, But Bet on Beliefs?" (2013)
  40. Buterin, V. "An Introduction to Futarchy" (2014)
  41. Wood, G. "Ethereum: A Secure Decentralised Generalised Transaction Ledger" (2021)
  42. Petrowski, J. "Polkadot Governance" (2020)
  43. Zill, B. "The Sybil Attack" (2002)
  44. Rodrigues, U. "Financial Contracting with the Crowd" (2019)
  45. Rodrigues, U. and Stegemoller, M. A., "Exit, Voice, and Reputation: The Evolution of SPACS" (2011)
  46. Rodrigues, U. "The Seductive Comparison of Shareholder and Civic Democracy" (2006)
  47. Warren, M. E. "Voting with Your Feet: Exit-Based Empowerment in Democratic Theory" (2011)
  48. Morgan, D. E. "Associative Democracy: decentralisation of societal and industrial governance? A critical discussion" (1996)
  49. Locke, J. "A Letter Concerning Toleration" (1689)
  50. Resnick, P. "RFC7282 On Consensus and Humming in the IETF" (2014)
  51. Clark, D. "A Cloudy Crystal Ball - Visions of the Future" (1992)
  52. de la Rouviere "The Moloch DAO: Collapsing the Firm" (2019)
  53. Steiner, J., & Dorff, R. "Decision by Interpretation: A New Concept for an Often Overlooked Decision Mode" (1980)
  54. Steinberg, R. "In the Shadow of Law or Power? Consensus-Based Bargaining and Outcomes in the Gatt/WTO" (2003)
  55. Barcott, B. "In Novel Approach to Fisheries, Fishermen Manage the Catch" (2011)
  56. Gutiérrez, N. L. and Hilborn, R. and Defeo, O. "Leadership, social capital and incentives promote successful fisheries" (2011)
  57. Orpheus Chamber Orchestra
  58. Sørensen, E. and Torfing, J. "Theories of Democratic Network Governance" (2007)
  59. Zurrer, R. "Keepers: Workers that Maintain Blockchain Networks" (2017)
  60. Zurrer, R. "Keepers Reboot – Observations & Forward Looking Trends" (2019)
  61. Unchained Podcast "CoinFund's Jake Brukhman and Multicoin's Tushar Jain on Generalized Mining" (2019)
  62. Burniske, C. "Productive Capital in Cryptonetworks" (2018)
  63. Daian, P. and Goldfeder, S. and Kell, T. and Li, Y. and Zhao, X. and Bentov, I and Breidenbach, L. and Juels, A. "Flash Boys 2.0: Frontrunning, Transaction Reordering, and Consensus Instability in Decentralized Exchanges" (2019)
  64. Qin, K. and Zhou, L. and Gervais, A. "Quantifying Blockchain Extractable Value: How dark is the forest?" (2021)
  65. Robinson, D. and Konstantopoulos, G. "Ethereum is a Dark Forest" (2020)
  66. Lloyd, W. F. "Two Lectures on the Checks to Population" (1833)
  67. Hardin, G. "The Tragedy of the Commons" (1968)
  68. Nakamoto, S. "Bitcoin: a peer-to-peer electronic cash system" (2009)
  69. Aligica, P. D. and Sterpan, I. "Governing the Fisheries: Insights from Elinor Ostrom's Work" (2017)
  70. Pennington, M. "Elinor Ostrom on Privatisation" (2012)
  71. DuPont, Q. "Experiments in Algorithmic Governance: A history and ethnography of "The DAO," a failed Decentralized Autonomous Organization" (2017)
  72. Zhu, S. and Hasu "Independent Property Rights" (2018)
  73. El-Hakim, S. "The Structure and Dynamics of Consensus Decision-Making" (1978)
  74. Lee, G. K. and Cole, R. E. "From a Firm-Based to a Community-Based Model of Knowledge Creation: The Case of the Linux Kernel Development" (2003)
  75. "KeeperDAO: A self-iterating liquiditing underwriting machine" (2020)