Research Summary: SOK: Blockchain Governance

TL;DR

  • No standardized framework for assessing the governance of blockchain protocols exists to this date. The aim of this paper is to use a survey of knowledge to provide a systematic account of blockchain governance and attempt a comparative assessment of major protocols.
  • The result is a standardized list of governance properties: Confidentiality, Verifiability, Pareto Efficiency, Accountability, Sustainability, Liveness and Suffrage. According to this list, ten major blockchain platforms are assessed and compared resulting in a table of relative satisfaction of these properties.
  • The authors conclude that these properties cannot be simultaneously maximized. Trade-offs among them might be necessary for maximally efficient governance. Some properties lead to opposing outcomes and it is up to each platform’s stakeholders to figure out which properties they want to maximize according to their goals and aspirations.

Core Research Question

How do the governance structures of different blockchain networks perform given a set of standard governance properties?

Citation

Kiayias, A., & Lazos, P. (2022). SoK: Blockchain Governance. arXiv preprint arXiv:2201.07188. https://arxiv.org/pdf/2201.07188.pdf

Background

  • Governance: Management of power within a system; answers the questions of who makes decisions and how along with how managers are held accountable.
  • On-Chain Governance: Governance structures for a blockchain-based network that utilize information stored and written on the ledger. This information usually includes smart-contract and token distribution data that is used in voting on decisions. Also includes changes to the software that are accepted or rejected by network nodes. Changes can include methods for collecting information, structuring incentives, collecting signals from ecosystem participants, and consensus mechanisms.
  • Off-Chain Governance: Governance structures for a blockchain-based network in which decision-making happens off the ledger. This often involves forum discussions, chats, physical interactions and other kinds of informal decision making.
  • Consensus: A decision made by group members who developed, and agreed to support a particular decision.
  • Consensus Protocol: A means of validating transactions between parties without presupposing trust or prior knowledge of each other and without the need of a central authority.
  • Decentralized Autonomous Organization (DAO): A public and permissionless organization that is governed programmatically.
  • Sybil attack: An attack on a network in which an attacker uses pseudonymous identities to gain a disproportionately large influence in the network.
  • Proof of Work (PoW): A blockchain consensus mechanism that validates transactions in a block by requiring participants to prove that they expended a certain amount of computing power for solving complex cryptographic problems.
  • Proof of Stake (PoS): A blockchain consensus mechanism in which nodes risk their native assets for the privilege and rewards associated with validating transactions in a block.
  • Skin in the Game: To have “skin in the game” is to have incurred risk (monetary or otherwise) by being involved in achieving a goal. In the phrase, “skin” refers to an investment (literal or figurative), and “game” is the metaphor for actions on the field of play under discussion. The aphorism is particularly common in business, finance, gambling, and politics.
  • Treasury: A collective reserve of funds an organization holds that can be allocated to various proposals and causes.
  • Approval Voting: In approval voting, there is usually a single winner who is approved by the majority. There are two main strands: in plurality voting, the winner is the candidate with the greatest number of votes, and in score voting, the one with the biggest score out of all candidates wins.
  • Instant Run-off: Voters in instant run-off elections can rank candidates in order of preference rather than choosing only one. This method is currently used in national elections in Australia, Ireland, India and Papua New Guinea.
  • Token-holder Voting: An on-chain governance mechanism where a voting outcome is determined by token allocation…
  • Conviction Voting: Computer-aided blockchain-native voting system where a voter can simultaneously entertain various proposals by allocating different amounts of shares to each. Holding a preference for longer, continuously grows the conviction. Withdrawals also happen continuously and are time-delayed, reducing the risk of last-minute vote changes.
  • Quadratic Voting: Quadratic voting is a collective decision-making process that allows citizens to vote based on how strongly they feel about an issue by allocating credits to votes.
  • Pareto Efficiency: Pareto efficiency is theoretically the most optimal way to allocate resources in a market economy. Making a change in a Pareto efficient system would mean that somebody would necessarily become worse off.
    • With respect to plotting Pareto efficient outcomes, welfare economists often use a so-called production possibility frontier (PPF) which is a curve showing the maximum possible outcome combination between two goods or services. All the optima of a function between two or more quantities, that are tangent to the PPF, are Pareto efficient, while every outcome that is below the curve is Pareto Inefficient.
    • In reality, a Pareto Efficient outcome is incredibly hard to achieve outside of a model or a very small market economy. A Pareto Efficient outcome would mean that everybody will be optimally satisfied.
  • Arrow’s Impossibility Theorem: In the context of voting systems as studied by Social Choice Theory, Arrow’s Impossibility Theorem has shown that when ≥3 alternatives exist, there are certain properties that cannot be satisfied by a ranked voting system. Pareto Efficiency is one of them.
    • According to the theorem, the fairest outcome would be possible if the conditions of Unanimity and Independence of Irrelevant Alternatives are satisfied.
    • Unanimity (if every voter prefers X over Y then in the final outcome we would expect X to beat Y) and Independence of Irrelevant Alternatives (if X beats Y but Z beats Y, removing Z would not change that X beats Y).
    • But if the first two are satisfied, there can exist a pivotal voter that is able to subvert the outcome of the vote. For example, in a case of 5 voters with 3 alternatives, if voters A and B rank X over Y and C and D rank Y over X, then the outcome fully depends on how voter E chooses to rank X and Y., In that case, E acts as a Dictator.
    • It is impossible to avoid the possibility of Dictatorship in a ranked voting system if one of the other two conditions is not discarded. This means that democratic ranked voting systems will either be inefficient or dictatorial.
  • Pseudonymity: The deliberate concealing of one’s identity in online social platforms.
  • Ethereum: a decentralized blockchain that implements automated smart contracts for payment and transaction-oriented architecture.
  • Catalyst: Cardano’s native proposal platform and DAO that promotes community innovation.
  • Polkadot: Polkadot is a proof-of-stake blockchain platform with mostly-on-chain governance.
  • Tezos: Decentralized open-source block-chain with smart contract capabilities. It employs a unique consensus protocol called liquid proof of stake. Its governance model allows Tezos to be hard-fork resistant.
  • Compound: An Ethereum-based decentralized borrowing and lending protocol that helps users earn interest for their tokens.
  • Uniswap: Ethereum-based decentralized crypto exchange that enables users to swap between any ERC-20 compatible tokens.
  • DAI: ERC-20 compatible algorithmic stable coin that uses smart contracts to keep its price as close to the US dollar as possible.
  • MakerDAO: An Ethereum-based DAO with an open-source codebase that employs a two-token system (DAI-MKR) to support a wide range of governance functions.
  • Dash: An open-source project (forked from Litecoin) and cryptocurrency focusing on transaction speed, ease of use and user privacy.
  • Decred: An alternative to Bitcoin which attempts to solve Bitcoin’s scalability problem. It operates as a DAO and maintains its own token and network which employs a hybrid PoW/PoS validation protocol.

Summary

  • The authors attempt to fill a very important gap in the literature by constructing a systematization of governance properties for assessing blockchain protocols.
  • The authors survey a great deal of literature on governance (beyond the blockchain) in order to create a systematic way of assessing the success of a blockchain governance. They define governance properties (such as Confidentiality) that the authors use as measures for the relative success of governance.
  • Given these conditions, no currently existing blockchain system is able to simultaneously satisfy all properties. Doing so would be impossible given that some of them are oppositional.

Method

  • The paper employs a Systematization-of-Knowledge (SoK) methodology. The authors conduct a comprehensive literature review by going through a great amount of both academic and non-academic (grey literature) resources on governance.
  • Academic resources consist mainly of research on election systems, resource allocation and political science. Non-academic resources consist of blockchain white papers, documentation and reports.
  • The authors distill the most important notions for assessing the success of blockchain governance and create a standardized list of properties that they apply to the assessment of various major blockchain protocols. They choose these protocols in virtue of the clarity of their documentation and their relative diversity so as to capture as much of the variety present in blockchain protocols.
  • The result is a comparative table illustrating the relative satisfaction of the identified governance properties by each of the chosen blockchain protocols.

Results

  • Seven main properties of blockchain governance have been identified: Confidentiality, Verifiability, Pareto Efficiency, Accountability, Sustainability, Liveness and Suffrage.
  • The dimension of Confidentiality is divided into three types: secrecy, pseudonymity and coercion resistance.
    • Secrecy is defined in terms of the ability of an algorithm to guess the input of a participant better than an adversary. Secrecy is mostly needed in off-chain governance settings and may even be undesirable.
    • Secrecy is often conflated with Pseudonymity but they are not the same. To satisfy Pseudonymity, a participant must be able to participate in the decision-making process without revealing their real-life identity as long as they have the ability to vote on proposals (e.g. by holding a certain amount of tokens).
    • Coercion Resistance is a stronger form of confidentiality defined in terms of the ability of a voter to deceive their adversary. If one has voted as they intended but is able to deceive the adversary that they voted as the adversary wanted, then coercion resistance is satisfied.
  • The standard version of Verifiability is called end-to-end and is satisfied when both the individual participant (Individual Verifiability) and independent observers (Universal Verifiability) can attest to their inputs being properly tallied and recorded on the blockchain.
  • Pareto Efficiency in the context of this paper is essentially a formal definition of the most efficient voting outcome when given a set of alternatives. The authors apply it to the context of approval voting and converting choices from ranked voting to approval voting.
    • There are ways to improve the Pareto efficiency of a system by giving all voters full and equal access to information about other voters’ preferences and adjusting the approval threshold according to these preferences.
  • Accountability is satisfied when participants are held responsible for any change they bring in a transparent way. There are different dimensions of accountability: vertical (to a higher up in a hierarchy) vs horizontal (to a peer in a network) and collective (joint responsibility of a network) vs individual (personal responsibility in a network).
    • Having “skin in the game”, is a very common strategy for ensuring individual accountability, usually by requiring participants to invest, lock, or stake tokens.
  • Sustainability can be achieved in regards to the development of proposals (sustainable development) and in terms of their enactment (sustainable participation). Both are important for the change and evolution of a blockchain protocol. This can be realized through various kinds of incentives (monetary, social, etc)
  • Liveness is the ability to act in a fast and agile manner in novel and urgent unanticipated circumstances. An example is MakerDAO which has implemented a shutdown function that returns the funds to the investors in case of an emergency (such as a hack).
  • In general, voting in blockchain goes beyond the classical “one person, one vote” system and there are many alternative elaborations of how voting rights can be distributed. The authors refer to this as Suffrage and there exist five main types: a) Identity-Based Suffrage, b) Token-Based Suffrage, c) Mining-Based Suffrage. A further distinction is made between d) Meritocratic and e) Universal Suffrage.
    • In Identity-based suffrage, participants are required to prove their unique (human) identity, while in Token-Based and Mining-Based Suffrage they have to prove that they hold a certain amount of tokens or have a certain amount of hashing power respectively.
    • In terms of the breadth of application, voting can be restricted to the most active contributors building the platform or it can be open to more indirect ways of support (token-holding, mining). The former corresponds to a Meritocratic approach while the latter to a more Universal approach.
  • The authors apply these properties for evaluating the governance structure of ten major blockchain platforms. Together they represent an overview of all the current approaches to blockchain governance.
  • They also create a table of comparisons illustrating their partial satisfaction or dissatisfaction of each property in which we can see the relations between each blockchain platform in terms of governance. There are various high-level takeaways such as the fact that coercion resistance is rarely accounted for and that there exists a trade-off between sustainability and liveness.

Table 1: Overview of the evaluations of each property against each of the chosen platforms.

  • Each platform has its own uniqueness and peculiarity but they can be categorized into four broad groupings: PoW systems (Bitcoin & Ethereum), PoS systems (Tezos & Polkadot), Ethereum-based systems (Compound, Uniswap, MakerDAO) and Treasury-based systems (Catalyst, Dash, Decred).

  • Proof of Work:

    • In Bitcoin the governance processes are off-chain and informal, using forums for communication. The Bitcoin Improvement Proposal (BIP) process is the primary governance mechanism of Bitcoin. Decision-making happens through a process of peer review and editing of proposals where successful proposals lead to either soft or hard forks.
    • Ethereum is similar to Bitcoin in the sense that it uses an Ethereum Improvement Proposal (EIP). It differs only in the sense that voting is not based on mining.
  • Proof of Stake:

    • Tezos, a next-generation PoS system, utilizes its own governance protocol Granada. Its governance process is described as “self-amending” because its software is able to both sense and execute an update at the end of each proposal voting period. A crucial requirement for submitting a proposal is that the underlying codebase can be compiled without errors so that it can be instantly deployed as a smart contract.
    • Direct participation is possible with 8,000 TEZ tokens and the whole voting process is composed of 5 discrete stages (cycles) spanning a two-week period. The 5 cycles are: 1) Proposal period, 2) Testing-vote period, 3) Testing period, 4) Promotion-vote period and 5) Adoption period. The first period uses approval voting while 2 & 4 use a super-majority-based voting system where at least 80% of the total sum of votes has to be in agreement.
    • Tezos is a generally transparent system but because of its high degree of automaticity and lengthiness of its voting process, liveness is compromised.
    • Another PoS system is Polkadot, whose governance is based on an elected council and its role is to propose or veto referendums. The council also implements a prime member who represents the default choice for members who are unable to vote.
    • Polkadot uses the Phragmen voting algorithm which consists of approving certain councilors and then redistributing tokens to vote again from a more limited set of candidates. A technical committee exists that can also propose emergency referendums (with approval from the council) that are enacted automatically. This greatly improves the platform’s liveness.
    • Proposals form the basis of future referendums and can be submitted by all stakeholders. General token-holders can vote by locking their DOT tokens (normal vote), not locking them (weaker vote), or keeping them locked longer than one enactment period (stronger vote). The choice of locking tokens greatly improves accountability, while the last choice makes Polkadot the only platform in this comparative study where Pareto efficiency is fully satisfied.
  • Ethereum-Based:

    • In Compound, Governance Processes are fueled by its token COMP which can be delegated to other addresses for gaining voting power. They can also be used for creating proposals.
    • An autonomous proposal can be made by any address with 100 COMP which can turn into a government proposal if 65,000 COMP end up being delegated to it. Two smart contracts control the governance process: Governor Bravo (2-day review, 3-day election period) and Timelock (2-day locking period which gives the opportunity to cancel a proposal before its final execution).
    • The governance process of Uniswap is almost Identical to Compound, although it ensures that voters are better informed about upcoming proposals through a Temperature Check and subsequent Consensus Check on the off-chain forum. A certain amount of UNI, its native token, is needed to go forth at each stage and reach the stage of a governance proposal.
    • MakerDAO: Combination of Uniswap’s off-chain and Compound’s on-chain governance.
    • Forum Signal Threads are used to get feedback on decisions before on-chain voting. They are followed by a poll where every user has one vote, irrespective of the amount of MKR they hold.
    • There are two on-chain processes facilitated by smart contracts: Governance Polls are used to get a sense of which proposals to enact, which can become Executive Votes, which are sets of code upgrade instructions intended to be executed once it wins the majority within a certain time.
    • The emergency shutdown feature provides great security and improves the protocol’s liveness while the on-chain voting process improves Pareto efficiency by avoiding a sequential vote.
  • Treasury systems function with a set budget that is considered when voting for proposals: Having budgeting in mind further complicates the governance process by providing an additional constraint in voting for proposals.

    • Project Catalyst is an on-chain decision-making process for how to distribute funds on the Cardano Blockchain.
    • There are four types of members: proposers, voters, Community Advisors (CAs) and Veteran Community Advisors (vCAs). The difference of Project Catalyst is with the CAs who review proposals and are rated according to the quality of their reviews by the vCAs. The voters then have enough information to vote on the proposer’s proposals.
    • The voting mechanism uses fuzzy threshold voting.
    • The great advantage of Project Catalyst compared to other platforms on the list is that it has a mechanism in place to ensure secrecy (and thus coercion resistance). Voters submit encrypted ballots and the committee decrypts them without revealing any information about the voter.
    • The relative disadvantage however is that the layered governance structure can be a bit centralized. For example, there is no voting mechanism in order for regular voters to become community advisors.
    • Dash is a PoW system similar to Bitcoin, but in Dash, the governance takes a formal on-chain form. Masternodes are an important part of Dash’s governance and are essentially nodes with stake in DASH big enough to be used as collateral for participating in governance. They each submit one approval vote.
    • Dash, like Catalyst, uses threshold voting but the percentage is higher (10%). Masternodes are rewarded for their participation and so sustainable participation is satisfied.
    • Decred: hybrid PoW and PoS system that is mostly on-chain governed. Governance rights are distributed as tickets for locking DCR tokens.
    • Governance has both on-chain and off-chain components. The off-chain component concerns proposals that require funds from the treasury (and uses a platform for getting a snapshot of tickets) while the on-chain component has to do with updating the software, which requires a substantial percentage to reach quorum.
    • Proposals compete for funding off-chain. The final (on-chain) vote is a referendum on already approved (off-chain) proposals.

Discussion and Key Takeaways

  • Trade-offs: Some deficiencies are necessary because some properties are in conflict with one another. Some trade-offs are:
    • Confidentiality - Verifiability: Confidentiality reduces the availability of information crucial for tallying and verification and verifiability is more difficult to achieve if the platform is optimized for privacy (secrecy, coercion-resistance) and thus there is a trade-off. But it is possible to achieve a good balance between the two.
    • Sustainable Participation - Accountability: Although these notions are closely related they can still be adversarial if rewarding and punishing are not addressed in an equal measure.
    • Token-Based Suffrage - Identity-Based Suffrage: In token voting (or coin-voting) votes correspond to tokens (meaning that one person can have multiple votes according to the amount of tokens that they hold) while in identity-based systems votes correspond to persons (meaning that one person, one vote).
    • Treasury systems have significant advantages in terms of sustainability by incentivizing development and participation through the use of treasury funds but liveness is seriously undermined by the long decision-making processes so as to guarantee security.
  • Room for Improvement: What the results point out is that there is no single best platform in terms of governance. All have significant deficiencies, which also means that they all have plenty of room for improvement. Through this standardized set of governance properties, it is clear exactly where these deficiencies are and what to improve upon.
  • Sophistication: The aforementioned set of governance properties provides a more nuanced assessment of blockchain platforms as it avoids the pitfalls of categorizing governance patterns as either centralized or decentralized by making no explicit reference to decentralization as a governance property. Various proposed properties could be proxies to decentralization but the current assessment is more concerned with the efficiency of governance rather than its structure.

Implications and Follow-ups

  • The authors hint at the fact that substantial improvements are needed in the platforms they analyze and that a design schema could be implemented according to the proposed standardized list of properties and their associated trade-offs
  • The authors, being both researchers at IOHK (Input-Output Hong Kong) which is the team behind the development of the Cardano blockchain, might be a bit biased in their overall assessment. From the resulting table of comparisons, Catalyst (the Cardano DAO) seems to outcompete the other platforms. The same is true for Polkadot which is closely tied to Cardano (since both their founders have a history of collaborating closely, mainly in the development of the early Etherem) which can be attested by the fact that both platforms satisfy properties that no other platforms do (Coercion Resistance, Accountability). In contrast, there is much less space allocated to talking about Ethereum governance which might be a suspicious move given the turbulent history of the Ethereum Founders.

Applicability

  • Debugging or updating the codebase of a blockchain system requires sound governance procedures for implementing changes in a timely manner. Standardizing governance properties will make the process of assessing the soundness of governance much easier, thus improving the quality of software and its usability.
  • A more standardized approach to assessing governance could help blockchain communities to audit their platforms and figure out the relative strengths or weaknesses of their governance.
6 Likes

How do you feel about the criteria? Do you think it’s an effective way of categorizing blockchain governance? Would you add other categories?

Governance is a crucial subject in any field involving human dealings. Using the Ethereum Blockchain fork as a case study, @Austin_jul’s research summary highlights the importance of blockchain governance.

The case of decentralized governance is different and harder to achieve because the system has to carry the community along. This stirs up problems when compared to centralized governance. The research summary by @windr shows one spectacular case of such governance problems where the community tries to game the system’s governance.

Bitcoin alone has had as much as five versions arising from hard forks of the original version. These hard forks were mostly a product of governance issues and disagreements.

What stands out for me in this present research summary is that, for once, governance on blockchain systems is considered from a different perspective other than centralization or decentralization as a governance property.

@Hermes_Corp Given that the seven criteria listed are somewhat oppositional and that one blockchain cannot possibly employ all the criteria, is there an optimum number of criteria a particular blockchain governance system must have to consider it the “ideal blockchain governance system”?

As this research made specific evaluations using certain blockchains as examples, these blockchains could consider the weaknesses highlighted from the evaluation. If they are considered, further experimentation and research can be carried out using the network’s testnet. If the results are confirmed to be positive, they can then be adopted and applied on the mainnet.

2 Likes