10.28.2020 | Fred Ehrsam
“THE BEST GOVERNMENT IS THAT WHICH GOVERNS LEAST” — UNKNOWN
Those who read my blockchain governance post from 2017 know I was excited about crypto exploring different forms of governance, especially on-chain governance. In the 3 years since, on-chain governance has gone from obscure to a popular component of many important protocols. Despite this traction, I think the most widely used protocols will trend towards governance minimization.
Why? Governance minimization allows stakeholders to depend on a protocol. This creates a virtuous cycle of adoption, enabling scale that would otherwise be unachievable. Look no further than successful traditional internet protocols like HTTP and SMTP to see this power today. Their governance minimization has allowed them to become standards everyone depends on, generating levels of use far exceeding what any company could achieve.
This post explains governance minimization, where governance remains important, and speculates on implications.
Governance minimization means reducing the power and reliance on governance wherever possible.
Governance minimization is important because it supports the primary value proposition of protocols: credible neutrality. Minimizing governance tends to make protocols more credibly neutral.
In practice today, governance minimization most directly applies to minimizing on-chain governance. If a protocol wants to achieve maximum adoption and can avoid using governance for a function, it should do so.
Credible neutrality is dependability. It means a stakeholder (e.g a user or a developer) can use or build on a protocol with confidence it will not change against their interests. Protocols remain credibly neutral by avoiding “capture” by any particular group.
Credible neutrality is primary value proposition of crypto today. It’s what allows for new forms of open and predictable money. It’s what creates platforms where anyone can openly create and access new applications.
Lack of credible neutrality in existing platforms makes its importance clear. Central banks are not credibly neutral as platforms for money because a subset of stakeholders can decide to print money at will. Facebook and Twitter are not credibly neutral as platforms for applications because a subset of stakeholders can make changes that kill entire developer ecosystems.
It creates safety of value locked in a platform — both concrete (funds) and abstract (development time, users) — from theft, shutdown, and restriction.
A total failure of credible neutrality is an exit scam. Some stakeholders of a platform can abscond, in part or in whole, with the value other stakeholders own or have created. A literal example is a centralized crypto exchange operator running off with user funds. An abstract example is Twitter shutting down third party developer interfaces to enshrine the interface of Twitter Inc.
Governance minimization is not the idea that governance can or should go away entirely. In fact, this is impossible! There are some protocol functions that are hard or impossible to perform without governance (“essential governance”, more on this later). And at a minimum, governance always exists through the ability to coordinate socially (“informally”) and hard fork.
Creating a perfect governance system is like designing a perpetual motion device. There are well-known game theoretic results which show all governance systems are both inherently unstable and fail to simultaneously satisfy all desirable properties. Formalizing a governance system tends to magnify these inherent issues by reducing the ability for informal governance to solve edge cases as they show up.
As a result, the more formal a governance system, the more susceptible it tends to be to capture over a sufficient time horizon. This is primarily because formalizing a governance system requires formalizing stakeholders (who has a say and how much of a say they should have). Formalizing stakeholders is very hard in practice at a single point in time and near impossible over time.
Formalizing stakeholders for on-chain governance today typically occurs through simple token ownership and the governance itself through tokenholder voting. Unfortunately, token holdings are not an accurate reflection of stakeholder value in a protocol for a number of reasons, including:
As a result, tokenholder voting creates a litany of misalignment issues with the long term use of a protocol, including:
Exploits to act on these misalignments are well documented and hard to avoid. Prominent attacks include buying votes or voting with borrowed tokens — an attack made easier with the invention of flash loans that has already worked!
Even if participants don’t explicitly capture a system, governance may still create unintended side effects. For example, formal governance systems can encourage action versus inaction, even when inaction is superior. Governance can also be extremely time and energy consuming.
A common argument for on-chain governance is that it can allow protocols to evolve faster. However, there is little, if any, empirical evidence of on-chain governance causing faster protocol evolution. On-chain governance is also extremely nascent, so judging it too heavily is likely premature.
What has been demonstrated is the opposite: that rapid evolution can occur without on-chain governance. For example, early Ethereum and Uniswap are protocols that evolved quickly through hard forking alone. This agility can be attributed to a combination of tight knit communities, lower systemic importance, and high levels of trust in core development teams.
What about ossification at later stages? Evolution may slow as some combination of network effects, pointer stickiness (protocols becoming embedded in the code of other protocols/applications), and risk aversion set in. Still, if a change is important enough, users and applications can voluntarily opt into it. This usually means a hard or soft fork for a layer-1 protocol and a voluntary migration for layer 2 protocols or application, as seen with the v1 to v2 transitions in protocols like Uniswap, Maker, Compound, and Augur.
Ultimately, if a protocol becomes sufficiently inferior to a new alternative, eventually it will be surpassed and die. This is how evolution works in nature. Individual organisms do not evolve (in the genetic code sense). Evolution occurs when copies are made with mutations. This approach has also worked in open source in the past (e.g Linux).
Since credible neutrality is the primary value proposition of protocols, erring towards governance minimization over time, even if it means slower evolution at the individual protocol level, will lead to maximal use.
Most importantly, dependable individual protocols enable faster ecosystem-wide evolution because developers can use these building blocks with confidence. Imagine how much faster development of internet applications would be if every company’s code was open source, neutral, and usable by anyone.
Governance is needed where a core mechanism of a protocol requires human input. Human input is needed when a protocol’s response to a particular action cannot be known in advance or derived from on-chain data, and thus cannot be coded into the protocol. We refer to these mechanisms as “essential governance”.
Areas of essential governance include:
Areas where governance is likely required include:
Some governance can be eliminated over time. For example, Maker could build a mechanism that programmatically sets interest rates, eliminating the need for governance to set them.
Governance is often a complex tradeoff. For example, the Maker system chooses to allow many different collateral types for the benefit of greater capital efficiency. This, of course, also has the drawback that the ability to handle new collateral types requires governance and all of the system risks, community energy, and complexity associated with that governance. It’s possible to build a governance minimized alternative to Maker with only one type of collateral; whether or not this system would be “better” is challenging to assess.
Finally, governance may be a feature in some cases. For example, people changing the rules of a game (governance) could be a core mechanic of some games or social apps.
While governance minimized protocols are more dependable, adverse outcomes for some subset of stakeholders can still occur. For example, the majority of participants in a governance minimized protocol can still choose to hard fork for their own benefit and at the expense of others. However, the cost is high: setting a social norm of violating credible neutrality causes future potential stakeholders to lose trust and eliminating stakeholders tends to reduce the network effects and use of a protocol. Forking can also be a feature, of course: it allows two different stakeholder groups to live on the protocols they each want.
Most importantly, eliminating governance in some places does not mean it can or will be eliminated everywhere. It may even make governance more important in the smaller number of areas it concentrates.
Well functioning governance remains vital, especially in areas of essential governance (e.g. layer 1 consensus, oracles). If the point of a blockchain is to provide a ledger of universally accepted truth, its integrity is paramount.
Innovation in on-chain governance systems remains an important frontier. Improvements beyond the current naïve “one token, one vote” standard likely exist (e.g. quadratic voting, futarchy) and are important to create. I remain excited to watch those experiments get run and continue to believe we’ll see more innovation in governance systems through blockchains than through “the real world” in the coming decades.
The extent to which governance is needed to capture value is an experiment playing out in real time. Pointer stickiness, gas costs, user-initiated state (e.g. a live bid in a decentralized exchange which a user would need to re-initiate in a new protocol), and other network effects are some factors that could create defensibility and allow a governance minimized protocol to capture value.
Value capture = use x take rate. The central claim of this post is that use of governance minimized protocols will be higher (the left side of the equation). Take rate (the right side of the equation) remains an open question. It may be that governance minimized protocols have higher use but lower take rates relative to a traditional company. And where that nets out economically is to be discovered.
Separately, many tend to hear “value capture” and think “evil!”. Value capture can serve a few important purposes. First, it incentivizes creation of protocols in the first place. Second, if governance is needed, governance value capture is necessary to incentivize contributions to governance and for security (to make 51% attacks on governance more costly). Finally, it can help fund further development.
Ultimately only “essential governance” is defensible. Anything inessential is likely to be competed away over time.
Governance minimized protocols will see the most use. It is the core attribute which kicks off the positive feedback loop between trust and adoption as a standard. It also puts powerful, basic tools in the hands of all creators, generating more opportunity and faster progress for the entire crypto ecosystem. As crypto grows into one of the most important technologies of the coming decades, it can lead to greater opportunity and faster progress society-wide.
Acknowledgements: Thanks to Brian Armstrong, Vitalik Buterin, Gus Coldebella, Jacob Horne, Matt Huang, Georgios Konstantopoulos, Steve Lee, Charlie Noyes, James Prestwich, Dan Romero, and David Vorick for conversations which contributed to this post.
Disclaimer: This post is for general information purposes only. It does not constitute investment advice or a recommendation or solicitation to buy or sell any investment and should not be used in the evaluation of the merits of making any investment decision. It should not be relied upon for accounting, legal or tax advice or investment recommendations. This post reflects the current opinions of the authors and is not made on behalf of Paradigm or its affiliates and does not necessarily reflect the opinions of Paradigm, its affiliates or individuals associated with Paradigm. The opinions reflected herein are subject to change without being updated.