The Collective Action Problem of On-Chain Governance
Defenders of on-chain governance praise the model for its efficient decision-making and rapid implementation. Critics of the on-chain model argue that it disenfranchises Node Operators removing the critical ability to fork a protocol update, or not. The defenders respond:
“Stakeholders have a very clear and broad incentive to do what’s right for the network”
“A stakeholders incentive to increase the value of existing token holdings obligates them to do what is best for the platform”
Who Are the Stakeholders?
While stakeholders include a myriad of different interests on the blockchain, they may be broadly categorized into one of four groups: Developers, Miners, Users, and Coin Holders.
Developers are the people writing the software that nodes will use to communicate with each other. There can be multiple, independent teams working on different implementations of the software. They benefit from:
1. Social recognition;
2. Legitimacy over the future direction of the software; and,
3. Increasing value of existing token holdings.
Node Operators are the people who run full nodes. They bundle the transactions along with transaction fees that are part of the throughput set. Their main source of income stems from the success and efficiency of the blockchain. They benefit from:
1. Expected future block rewards;
2. Expected future transaction fees; and,
3. Increasing value of existing token holdings.
Users are the people who find efficacy in the decentralized application platform. They transact and exchange goods and services on the platform. Their stake is decided by their usage of the platform itself. They benefit from:
1. Increased functionality;
2. Faster transaction speeds; and,
3. Influencing the value of existing token holdings.
Coin Holders are the people who own a significant amount of tokens on the blockchain. They sell their tokens to Users, who then utilize them for goods and services on the platform. In Proof-of-Stake systems, Coin Holders may also purchase token as a store-of-value (I will return to why this is important later). They benefit from:
1. Stake-weighted voting systems;
2. Increasing their stake on the blockchain; and,
3. Increasing the value of existing token holdings.
So here is the problem. Yes, all of the stakeholders as defined above benefit from influencing the value of existing token holdings. But for each of the groups, affecting the value of the token is only 1/3 of their overall incentives for the network. This means that to falsely assume all stakeholders want what will increase the value of the tokens implies: 1) that is their top priority for the entire network, and 2) in a situation where a stakeholder must choose between token value or another benefit, token value will always be paramount.
The result of this flawed logic is the assumption that the only incentive for stakeholders is token value. To sum up this fallacy:
Persons using the blockchain network may only be considered a stakeholder if, and only if, his sole priority is increasing the value of existing token holdings.
Power Distribution in On-Chain Governance
Ideally, the four interest groups have a population distribution resembling a pyramid. Developers are the smallest interest group, Node Operators the second, and so on. But in off-chain governance, each of the interest groups is given an equal amount of power.
In off-chain governance, the equal power distribution of the model allows for tension between the separate interests and incentives of each group. When each group has an even distribution of power, it keeps one group from dominating over the other. Maintaining a balance of power between different interests allows each one to feel enfranchised.
In evenly distributed systems, Coin Holders have an equal impact on the direction of the network compared to other groups. In stake weighted systems however, power moves to the hands of Coin Holders; there is somewhat of an oligopoly of coin owners and a majority of the wealth is held by a small percentage of the participants. For example, 1.6% of EOS holders own 90% of the tokens. If the voting system is based on stake, then only 1.6% of the participants are going to have any sort of a voice.
In this way, on-chain governance has reinstated the pyramid structure as a power distribution. While someone may be both a coin holder and a user, or even if most users are coin holders, on-chain models that engage in stake-weighted voting unfairly skew power towards coin holders. Even in networks with evenly distributed wealth, the system empowers the Users and largely disenfranchises the Node Operators and Developers who play such a critical role in the environment of the blockchain.
Why the Blockchain Community Should Care
Say an on-chain entity was hacked, and US$40 million was siphoned off of a smart contract. A Proposal is on the table to create an irregular transition to restore the stolen tokens to their original owners. Unfortunately, the oligarchy of current token holders at the present block yield more power now than they would if the stolen tokens were returned. Therefore, these oligarchs are not going to vote to return the stolen tokens and release their increased power on the blockchain. The larger group of Users, with a smaller stake, will not have their lost tokens restored and token-whales will continue to hold a majority stake. As a result, the disenfranchised Users may choose to leave the network.
Keeping with that example, imagine the same US$40 million was stolen through a hacked smart contract that Developers had invested in. In this instance, no token-whales exist in the space. A Proposal is on the table to create an irregular transition to restore the stolen tokens to their original owners. Unfortunately, in this instance the Developers are not going to hold a significant stake to vote in support of the Proposal, and most Users have no incentive to vote in support because their tokens were not stolen.
This is the major fallacy in the argument for on-chain governance. It fails to recognize these inherent issues among stakeholders. In politics, we call this a collective action problem. It is a situation in which an individual will make a choice that is best for his or her self-interest, even if the alternative is better on the whole for the entire group. If each individual follows their self-interest, the outcome is generally worse for everyone than if all had cooperated in supplying the public good.
Now, had this Proposition been posed in an off-chain model, token-whales, even if they exist, wouldn’t have power over the direction of the governing model. The majority of Users would have a voice to shift the direction of the chain after the hack, and the process would be overall much more democratic. In the second example, Developers would exercise their power as equal stakeholders against Users, rather than minority stakeholders in on-chain voting, to vote on the proposal. None of the groups would feel the need to circumvent the formal legislative process because each voice would be fairly heard.
Why Off-Chain Governance?
On-chain governance automatically updates once the voting threshold is met for any single proposal. This necessarily means that Node Operators are debilitated in their ability to decide to install the update, or not. If the dissemination of wealth is uneven, and these token-whales rule the space, this also necessarily means that both Users and Node Operators are disenfranchised.
A Node Operators conscious decision to install a hard fork, or not, provides the balance of power necessary to create tension between interests groups, as depicted previously. If Node Operators and Users follow the on-chain process, they’re naturally disenfranchised, but more importantly, increasingly reluctant to refuse a hard fork update. To top it off, Vlad Zamfir notably wrote that on-chain voting is not Sybil-resistant; any full node participant will, by extension, be robbed of their modest participation in a model already working against them.
In off-chain governance: yes. Yes, it is possible for one group to dominate over the other. Yes, collective action problems exist. Yes, wealth dissemination may be uneven. The difference is that 50% + 1 off-chain is not a revolution as it is on-chain.
Off-chain, Developers will express their hopes for the direction of the platform, and generally play a significant role in the development of hard fork updates to the protocol, at least in the early life of the blockchain platform. Node Operators will engage in discussion and express their concerns for the update and, at the end of the day, make their final decision by updating the protocol, or not. Users and Coin Holders vote, everyday, when they make a conscious decision to continue engaging and investing in the platform.
Their factional identities are independent of each other. Expectations from the platform operates as a two way street: Developers, Node Operators, Users, and Coin Holders expect different things out of the blockchain, and the platform, as an organism, expects different things from each of them as well. The social hierarchy provides for an ecosystem that can thrive.
These are human issues, not blockchain issues nor cryptocurrency issues. Both models of governance are culpable to these types of collective action problems. The difference is that off-chain governance provides the flexibility to adapt and change to negate collective action problems in a much more fluid way than on-chain models do.
Conclusion
By this point, it’s clear that incentives between Developers, Miners, Users, and Coin Holders are perpendicular to each other in many potential governing issues. So as a stakeholder group, these four sub-groups are going to have conflicting solutions to each problem stakeholders face.
Zamfir made a poignant statement on the topic:
The information and incentives of participants constrains their participation, and they need to be understood in the context of the underlying culture, as well as their individual contexts. The information and incentives of participants can change over time, but they don’t change instantly, nor do they change independently of each other’s incentives and states of knowledge. The process of changing the way that something is governed is not magic, and in our case it’s also very human.
Governance models that succeed will be versatile enough to change and adapt over time. All constitutions should be considered living documents with the ability to grow. In blockchain, the result is an imperfect system. But, it balances the platform by allowing for tension between each faction making tyranny of the minority virtually impossible. On-chain governance is not democracy, it’s tyranny.
