Study on the Quadratic Voting Grant Model (1)

Trias
8 min readJun 9, 2021

On May 15, DAOeco initiated the Trias Grant #1 Quadratic Voting with the aim of designing a better voting scheme. Now it has some progress.

The following is the first part of the academic report.

The research of DAOecoe is to design a model of Quadratic Voting for Trias Grant. Compared with one person, one vote, and one coin, one vote, Quadratic voting is a relatively fair voting method that expands the influence of the ordinary and weakens the control of the rich over the community, allowing more people to benefit from community governance.

Focusing on this research topic, DAOeco initiated discussions on Quadratic Voting in the community and invited Yuqing Hu, who has worked for the World Bank with a master’s degree in economics from Duke University and also has studied under the Chinese interpreter of Radical Markets to shared about Quadratic Voting and liberal activism. The discussion has played an important role in this research.

Based on Yuqing Hu’s sharing and related papers, DAOeco proposed a more decentralized way to prevent the Sybil Attack in Quadratic Voting in response to the anti-cheating challenge, making the voting results fairer and in line with the wishes of the community.

Background

In the governance practice of DAO, voting mechanisms and algorithms are very important. In 2018, the paper Liberal Radicalism: Formal Rules for a Society Neutral among Communities published by Vitalik Buterin in collaboration with Harvard scholar Zoe Hitzig and Microsoft’s E. Glen Weyl discussed how to establish a reasonable voting system for public fundraising projects. Related to the paper, this article aims to discuss the advantages of Quadratic Voting and the challenges in reality.

The theme of Liberal Radicalism: Formal Rules for a Society Neutral among Communities is to propose a decentralized and self-organizing ecosystem that uses seed funds raised publicly to allocate public goods.

Public goods are a classification that opposes private goods in economics, and it has two characteristics:

1. Non-competitive: The consumption of a public item by some people will not affect the consumption of the item by other people. For example, national defense protects all citizens, and the cost and benefits of individual citizens from the national defense will not change because of the newborn or the departure of a citizen;

2. Non-exclusive: Citizens don’t have exclusive rights to public goods. For example, the government purifies water so that citizens can drink clean tap water. It is impossible to prevent anyone from using purified water.

In fact, when the society allocates public goods like infrastructure, public education, national defense construction, etc., the two most basic problems are the free-riding problem and the majority overwhelming minority interests.

Free-riding problem: The non-exclusive and non-competitive nature of public goods lead some people to declare in advance that they do not need a certain type of public goods, and after others have paid for it, they can enjoy this public product for free. In a system, the more free riders appear, the more insufficient the supply of public goods will be, and the market will gradually fail as a result. For example, under the planned economic system, it is commonly thought that to work and not work are the same, which led to serious internal friction effects and inactivated economy obviously.

The majority overwhelming minority interests: When a system is completely based on the one-person-one-vote model, the system cannot reflect the importance of diversity of products, thereby suppressing minority’s interests. For example, diabetic patients are a minority of the total population. Under the strict one-person-one-vote model, it is difficult for them to put forward proposals on their demands for insulin, let alone vote for such proposals.

In reality, public goods are mainly provided through three methods: government funding, private owner investment, and charity. Each of these methods has some drawbacks, and cannot effectively solve the above two basic problems.

In addition, the one-person-one-vote voting method itself has loopholes: under the assumption of rational people, when the number of participants in the election is large, a single ballot has a negligible effect on the result, so rational people may not vote, which causes free-riding in the vote and irrational results appear.

Another drawback is that the 1-unit-1-vote rule, representing the capitalist system, which means the one with more chips can speak louder, can not escape from the above problems.

The majority overwhelming minority interests exist in both government funding and private owner investment. Compared with the above two methods, charity and public welfare are more flexible and responsive, but it is based on the kindness of citizens. Funds rely on limited donations by them, so the supply of public goods may not meet the demand, which is a problem. Additionally, unless there is a centralized regulatory agency to charge over it, charities are prone to a free-riding problem.

The free-riding problem is commonly found in the social system. At present, there are three methods to solve it:

1. Taxation and democratic voting: In so-called democratic countries, voting and taxation are used to determine self-help items and amounts;

2. Privatization: Allow profit-oriented companies to provide public goods, and restrict those who don’t pay from enjoying them;

3. Rely on a sense of identity from morality, culture, and religion.

However, there are various drawbacks to the above mechanisms. The most prominent one is a free-riding problem. These mechanisms more or less rely on centralized regulation and coercion, or irrational assumptions about individual behavior. But this paper can naturally solve the free-riding problem through effective design of the incentive mechanism without relying on the above strong assumptions.

The basic concepts of Quadric Voting

The Liberal Radicalism model proposed in the paper Liberal Radicalism: Formal Rules for a Society Neutral among Communities is based on Quadratic Voting (QV).

QV is the first voting method proposed by E. Glen Wyel and Eric Posner in 2013. Different from the one-person-one-vote or one-chip-one-vote model, QV allows individuals to buy tickets, but they need to pay the square of the number of tickets they buy. As is shown in follows:

QV has obviously weakened the voting power of capitalists with more chips: originally they could represent 9 votes with 9 chips, but now they can only represent 3 votes with 9 chips. On the other hand, it also effectively increases the cost of monopolistic voting by buying other’s chips.

Another advantage of QA is that it can encourage citizens with a small number of chips to vote, because their voting power is magnified in QA, and they have more say than before.

The anti-cheating challenge of Quadratic Voting

Making a system fairer though, Quadratic Voting also has a fatal flaw that it is easy to cheat. People can get more benefits by dividing their voting costs by establishing clones and votes separately, which is also called a Sybil attack.

Quadratic Voting gives more weight to disadvantaged groups, but at the same time makes Sybil attack possible. Taking the voting scenario as an example, the final votes of a person from casting N votes will be less than a person employing N individuals and each casting 1 vote.

We have proposed several methods focusing on preventing cheating in Quadratic Voting in different scenarios and purposes.

1) Increase the cost of Sybil attack through staking

In short, the real on-chain ID is your asset. We can create a smart contract and require voters to put a certain amount of token into it before voting. After the voting, the smart contract will return the staked token to users.

For example, if we require 10 BNB for each account to be eligible to vote and someone wants to create 10 voting addresses, they need to stake 100 BNB in total, which increases the barrier to carry out Sybil attack.

2) Use centralized ID

The easiest way to solve this problem may be the centralized mechanism. For example, GitCoin grants require users to log in with GitHub, and in this way, GitHub automatically provides an identity for voting.

One of the differences between HackerLink and GitCoin is that HackerLink Grant runs on smart contracts instead of multi-signature wallets. Consequently, even if HackerLink requires users to log in from the front end, scientists can still interact directly with smart contracts.

Therefore, we can draw a conclusion that the current decentralized identity solutions (DID) are not mature enough in general, and increase the user’s usage cost, so that cannot effectively solve the problem.

In the GitCoin GR9 that just held, even if a centralized identity system is deployed, GitCoin still suffered from Sybil attacks. Therefore, we can say that a centralized solution is not the final one.

3) Deploy the whitelist to register as a voter

Whitelisting may be the most reliable method when voting on the chain. If voters are whitelisted before the voting, Sybil attacks will be automatically prevented. The disadvantage is that voters must register an account even before knowing the project, which will naturally reduce the number of people participating in donations and voting.

4)Progressive Tax

Another way to suppress Sybil attacks is to use a progressive tax (fee) system. More community donations will result in higher taxes (costs). For example, if the community donation for a single project exceeds 200 BNB, a 40% fee will be charged, leading to a much higher cost of the Sybil attack. The problem with this approach is that when donations exceed a certain level, the number of votes will be depressed. But if we can estimate the largest community donation amount for any project, a reasonable progressive tax system will be designed.

Ironically, since we usually conduct Quadratic Voting openly, it is difficult to estimate the maximum donation amount from the beginning.

5) Grace period

Due to the fierce competition among projects in the last few days, we will add stricter grace periods in future events.

During the grace period, there are two things that need to be done:

A. Verify project identity

B. Analyze voting records in the event and detect Sybil attack

The grace period allows us to identify crooks and projects that carry out Sybil attacks and cancel their right to obtain matching funds. For fraudulent projects, we must return community donations to users.

6) Use NFT incentives

NFT incentive is an improved method based on the whitelist mechanism. By issuing NFT rewards to past participants, we can link their addresses to the NFTs, each of which is equivalent to a unique identity. And the addresses will automatically enter into Grant’s whitelist. The disadvantage of this method is that it gives up the anonymity of the blockchain to a certain extent.

--

--

Trias

Trustworthy and Reliable Intelligent Autonomous Systems