On a decentralized platform, how do you decide who is in charge and what protocol changes are rolled out? In the blockchain world, many teams attempt to answer these questions through on-chain governance; they use cryptoeconomic models that allow people to stake their crypto to vote for changes they want to see.
This week at San Francisco Blockchain Week, leaders from 0x, Tezos, MakerDAO, and Polkadot shared their approaches in a panel discussion titled, quite simply, "Governance" (albeit the title was somewhat misleading, as the panel was specific to on-chain governance).
Some believe governance is dangerous and that, once launched, blockchain protocols should be left alone as much as possible. If code is law, and that's that, then you don't risk centralization of decision-making. This is crypto anarchism.
But you don't have to be a crypto anarchist to have doubts about on-chain governance. Vlad Zamfir famously (or about as "famously" as you can do anything in blockchain outside of being bitcoin) wrote a blog post last year expressing his deep skepticism about the legitimacy of designing governance into a blockchain protocol. (Zamfir was talking about layer one blockchain protocols and specifically stated that he was pro-experimentation with smart contract-level governance mechanisms, but many of his concerns apply to smart contract-based protocols.) His reasons were many, but included the issue of plutocracy, as well as ensuring voters have the necessary information to make educated decisions.
No one on this panel seemed to be a crypto anarchist or share Zamfir's skepticism of on-chain governance. As 0x co-founder and CEO Will Warren explained, upgradeability is crucial, and governance is required to make that happen. Ethereum, the blockchain 0x runs on, is still evolving at a rapid pace. If a team wants its protocol to remain useful, organizations running businesses on top of it must update their code with it. Warren pointed to new coin standards and languages as examples of these kinds of changes and said that if you don't have a system for upgrades, then users will leave for new protocols. This isn't just a problem for businesses – it's a lot of friction for users, too. It's better if users can instead have a say in changing the existing network to be what it needs to be. On-chain governance has the advantage of not requiring a third party to facilitate those changes.
There are several complications, though, when it comes to actualizing on-chain, tokenized governance, many of which are overlapping. These complications include identity verification, bribery, and token distribution.
One standout issue at the Governance panel at SF Blockchain Week was token distribution and the problem of plutocracy. At its most basic, the problem is this: In a cryptoeconomic governance model, fairness can be difficult to achieve because crypto buys decisions. You're literally asking people to pay for votes or, at least, risk losing crypto if they bet on the losing side.
This model, though, is in some respects quite utilitarian. The idea is that people who care more will pay more, meaning that in the end, the side who has the most "at stake" in the matter will also quite literally have the most crypto staked in the vote. Assuming that the "people who care more" are the same people who will be most affected, the decision reached under this model should then be the option that creates the most good (if not necessarily for the most people).
One can argue, though, that this is only the case if token distribution is such that most people have relatively equal overall voting power. Otherwise, it only means that whoever can afford to vote will win it.
One could also argue that those who can afford to vote stand to lose the most if things go south and, therefore, should be the ones making the decisions because they are most likely to value the health of the network. But loss can also be understood relatively: The loss of a small amount of value may have a bigger impact on one token holder than the loss of a large amount does on a richer token holder. The problem of whether or not stakeholders truly have a say is made worse if there is low voter participation, which is likely if token holders don't believe their vote(s) will matter.
This is something that Jacob Arluck, a researcher at Tezos, echoed. He argued that token holders won't participate if they believe it's just investors who are making all of the decisions, rendering the governance illegitimate. Arluck referenced a more recent "Governance 101" blog post by Zamfir to claim that for a governance mechanism to be legitimate, then people who are assumed to be participating (in this case token holders) should hold the reasonable belief that others like them will use the decision-making process or coordination mechanism. Arluck's reference to Zamfir's thoughts on blockchain governance was somewhat surprising, as Tezos is doing exactly what Zamfir warned against: experimenting with on-chain governance on the blockchain protocol level.
The Tezos network has a pretty extreme case of wealth inequality: The vast majority of tokens are held by a small few. This problem is exaggerated because one must hold a lot of Tezzies to participate in a vote in the first place. Tezos' way of solving this problem of legitimacy is to allow token holders to delegate their votes to what's called a baker, who will vote on their behalf. However, as of July, about half of all tokens used to vote were owned by the Tezos Foundation, and token holders who own only small amounts of Tezzies are least likely to vote. While Tezos expects participation to improve over time, the system might not be entirely broken (at least by one measure): As it is, the people who stand to lose the most if the token fails are the ones participating.
Identify Verification as a Solution?
Perhaps the most obvious solution to this problem is to eliminate cryptoeconomics from this equation entirely, and just structure the system so that each participant has one vote. The problem with this is identity verification. As it stands, blockchains and organizations running on top of them often lack sound identity verification mechanisms. This is rooted in a common ethos of privacy among folks in the blockchain community – and also a distaste for traditional banking and government systems. A lot of blockchain projects do not require government-issued or verified identification. Because of this, it's incredibly difficult to enforce one-person, one-vote.
The issue of identity verification came up during the panel, but no one seemed to be convinced that it was necessary – or, at least, they didn't think the tradeoff was worth it, although Robert Habermeier, the co-founder of Polkadot, seemed to lament the lack of adequate identity verification mechanisms, as did Steven Becker, the president and COO of MakerDAO. Becker argued that until users own their identities, you can't ask them to use identity verification methods on a platform.
Warren wasn't convinced that identity verification would improve things anyway, arguing that representatives (such that may exist in liquid democracy, proof-of-stake voting pools, or similar schemas) should be able to elect to provide whatever information they deem relevant. Then, voters can decide whether to trust the representative with their votes. Arluck admitted that the existing voting mechanisms were not great, but echoed Warren that adequate solutions don't necessarily require identity verification.
If you don't have identity verification, and you don't have a relatively even distribution of wealth, then maybe the best you can do is manage plutocracy.
Becker made no attempt to deny plutocracy within MakerDAO, but pointed to its complicated "Governance Risk Framework" to explain how it maintains Dai's stability despite wealth inequality. And Becker's point is fair: The Dai is stable. Though, of course, choosing to manage plutocracy rather than seeking to eliminate it means that the DAO isn't quite as decentralized as one might hope.
Habermeier emphasized the need for accountability among representatives/staking pools. If representatives are held accountable, this might incentivize token holders to participate in votes. For this reason, he pointed to futarchy, and the ability to easily boot representatives from their position, as a solution.
By contrast, Arluck pointed to liquid democracy as a way of minimizing or eliminating class. He argued that the mechanism will be legitimate if participants are incentivized to vote for delegates to vote on their behalf by earning interest off their staked tokens.
Warren raised doubts about the merits of liquid democracy. He credited Peter Zeitz, who recently joined 0x's core team as a researcher, for changing his mind, and pointed to low voter participation as a barrier to true liquid democracy. He used EOS as an example that it doesn't work. Instead, he pointed to more traditional representative democracy models.
Unsurprisingly, the panel did not solve all the problems of on-chain governance. In the end, no one has the answers. But they're asking the right questions.