As the decentralized finance (DeFi) market evolves, it seems paramount that new and current protocols incorporate more sophisticated governance techniques.
Governance is one of the fundamental building blocks of decentralized finance (DeFi). The ideas of decentralizing and democratizing governance in DeFi protocols are incredibly novel, but the techniques haven’t evolved at the same pace as the rest of the DeFi ecosystem.
Jesus Rodriguez is CTO and co-founder of blockchain data platform IntoTheBlock, as well as chief scientist of AI firm Invector Labs and an active investor, speaker and author in crypto and artificial intelligence.
For all the innovation that we have seen in the DeFi space, most protocols today are governed by decentralized autonomous organizations (DAO) that operate using a basic one-token, one-vote model. The simplicity of that model makes it incredibly easy to implement but also creates broad vulnerability surfaces for DeFi protocols. Just as we were writing this article, stablecoin protocol BeanStalk was a victim of a governance attack that resulted in a $182M exploit. This type of attack reflects the fragility of the one-token, one-vote model which, by definition, favors large token holders at the expense of other participants in the protocol.
Fortunately, there is a relatively unknown area of computer science known as computational voting theory that includes plenty of ideas that can be incorporated into next generation DeFi governance models.
A brief history of computational voting theory
Voting theory is an ancient area of mathematics tracing back to Plato’s unfinished book “The Laws” in which the Greek philosopher proposed multi-stage voting as an alternative to the traditional democratic models.
The golden era of voting theory took place during the Enlightenment and the French Revolution, when dogmatic ideas were challenged with science and math. During those years, thinkers Jean Antoine Nicolas de Caritat, Marquis de Condorcet, and Jean-Charles, Chevalier de Borda created methods such as Condorcet criterion or the Borda count, respectively, which have become the foundation of many modern voting systems.
During the 1950s, voting theory received new momentum with the work of economists such as Nobel Prize winner Kenneth Arrow. Arrow’s impossibility theorem has become the cornerstone of social choice theory.
The emergence of multi-agent systems in areas like artificial intelligence (AI) triggered a new wave of research in voting theory. The blockchain space has been an adopter of many novel ideas in voting theory in the form of consensus protocols. Although DeFi has been slow to adopt more sophisticated voting paradigms, we are already seeing modern ideas from computational voting theory making its way into DeFi governance models.
Beyond the 'one token, one vote' Model
Native tokens in DeFi protocols align users with an economic interest in the protocol while also giving them the right to participate in the governance process. This duality of value seems logical but becomes problematic in simpler governance methods like "one-token, one-vote" because large token holders can manipulate proposals to benefit their economic interest.
In general terms, there are two fundamental types of approaches to improve DeFi governance models:
- Build token models that decouple the economic interest from the governance participation. Removing this duality can indirectly lower the risk of market manipulation attacks.
- Implement more sophisticated voting models that require considerably more resources to manipulate.
There are plenty of ideas in modern voting theory that can be adapted to address these two key points.
1. Quadratic voting
One of the favorite options of the crypto community is quadratic voting (QV). The concept was originally proposed by scientists William Vickrey, Edward H. Clarke, and Theodore Groves but was really popularized by Microsoft Research New England’s economist Glen Weyl. The core QV model enables voters to purchase votes where the price of votes is a quadratic function of the number of votes purchased. When applied to DeFi governance, we can envision that voters will have to spend governance tokens at a cost that grows quadratically relative to the number of votes. Two votes will cost four tokens, three votes will cost nine tokens, etc.
Ideas like QV makes it harder for large token holders to manipulate governance proposals given the increased cost of acquiring votes. Gitcoin has been one of the most prominent projects experimenting with ideas around QV with relatively good results.
2. Holographic voting
A challenge for the "one token, one vote" model in DeFi governance is that all proposals receive the same level of attention regardless of their relevance. This allows large token holders to manipulate important governance decisions since they don’t get the proper attention of the community.
Proposed by Matan Field and several contributors from DAOstack, holographic consensus (HC) tries to curate attention to relevant governance proposals. The core mechanism uses tokens that can be staked to boost attention to relevance proposals. HC also introduces a second token that can be staked to bet on the outcome of a proposal. Token holders that bet on the right outcome get compensated while those voting against will lose their stake. The dual token mechanism drives attention to the important proposals while also ensuring that the economic interest of large token holders is aligned with the community.
3. Proof of participation
An interesting idea that has been outlined by several forward thinkers in the DeFi ecosystem, including Vitalik Buterin, is to limit governance to accounts that are actively participating in the protocol. Imagine a DeFi lending protocol that restricts its governance mechanics to addresses that have actively issued and borrowed loans, have a trading history without liquidations or have even contributed to different governance proposals. Although not bulletproof, this method can significantly limit governance attacks as voters will need to be actively involved in the protocol.
4. Limited governance
A somewhat obvious concept, also outlined by Vitalik Buterin in a recent blog post, is the idea of constraining governance proposals to certain aspects or parameters of a DeFi protocol. For instance, consider an automated market maker (AMM) protocol in which governance proposals are constrained to changing weights within certain ranges in liquidity pools. Similarly, DeFi protocols can introduce time delays between the time an impactful proposal is approved and its implementation, giving the community time to reflect on it and prepare for the impact. This should lower the probability of catastrophic attacks via governance manipulations.
Improved DAO-DeFi governance
The rapid level of innovation and growth in DeFi has challenged some the simplicity of governance models. The current generation of governance models represent an important vector of attacks and market manipulations. Just like voting models have evolved throughout our socioeconomic history, the next phase of DeFi requires more robust governance mechanisms. Thankfully, the answers might already be in the rich history of computational voting theory.