Emerging Designs for Collective Governance
Emerging Designs for Collective Governance
Part 1: Voting, Reputation and Decision-making
This article highlights the rise of new collective action methods that are being developed in, or are applicable to, the Web 3.0 and Platform Cooperativism space.
This article is divided into two parts. Part 1 covers voting, reputation, and decision-making methods. Part 2 covers organization technology (OrgTech).
1.1. Quadratic Voting
Quadratic voting “is a method of collective decision-making in which a participant votes not just for or against an issue, but also expresses how strongly they feel about it” .
A basic formula of the voting cost model for Quadratic Voting can be described as:
Voting cost = (Number of Votes)² 
In other words, the more a vote concentrates their votes on an issue, the fewer votes a voter gets overall .
The goal of Quadratic Voting is to improve upon the well-known and practiced Yes/No voting mechanism (i.e., Binary choices), which does not truly capture the preference of voters on all issues because of its binary nature (all or nothing regardless of varied preferences on multiple issues) .
With this type of voting mechanism, even if a voter has varied preferences, they are forced to concentrate their votes onto a single issue (usually the most important to them), even if it means they must vote against their preferences on other issues of which they only have a moderate preference for .
Since most of us are, rarely if ever, totally for or against a decision on an issue, Quadratic Voting seeks to allow voters to express their varied preferences on issues, and “‘speak louder’ on the issues they deem most important” .
Additionally, other than expressing voter’s varied preferences, Quadratic Voting also aims to mitigate against the “loudest voices in the room problem.”
Voters start out with equal voting credits (e.g., 10 voting credits), that they may allocate on issues up for a vote . For each vote on an issue, the number of credits spent is the number of votes squared, as shown in the formula above .
For example, if there are 10 issues up for a vote, and Jack chooses to allocate 2 votes each on Issues 1 and 2, then Jack’s allocated votes are 4, and his voting cost is 8 . Now Jack can only allocate 1 vote each on two other issues, because he only has 2 voting credits left .
If Jack felt particularly strong about Issues 1 and 2, then him allocating his votes for 1 and 2 would make sense given the voting costs, but if Jack did not feel particularly strong about Issues 1 and 2, and instead felt even stronger about Issue 3, then Jack has overspent his voting credits on these two issues .
Thus, if Jack did not have a strong preference for Issues 1 and 2, he has been penalized for concentrating his votes on Issues 1 and 2 .
On the other hand, if Jack did have a strong preference for issues 1 and 2, he has shown his preference and
Now if Jack was moderate on all Issues, Jack may simply allocate one vote for each Issue and show that he has a moderate preference on all Issues .
1.2. Conviction Voting
Conviction voting is a collective decision-making mechanism wherein all proposals in a community are considered at every point in time .
In this mechanism, member preferences for proposals is determined by the amount of preference for the proposal, and the longer a member holds a preference for the proposal, the greater the amount of conviction the proposal will have based on a “half life decay curve” up-to a certain limit, (i.e., positive relationship between time preference is held and the amount of conviction) .
At any time, the member may remove their preference for a proposal and switch it to another proposal, which removes the amount of conviction based on the half life decay curve .
Each proposal has a required amount of conviction for passage, thus we know when the proposal is inline with the preferences of the community .
Sourcecred is developing a reputation protocol (also known as social algorithm) for contributions to open source software projects . Contributions include anything that supports a project .
By mapping the relationships of contributions (“contribution graph”), Sourcecred plans to reward contributors for their contributions in cred by utilizing a PageRank algorithm to the Contribution Graph to generate a score . In Sourcecred, “a contribution earns cred if it is connected to other contributions that earn lots of cred” .
The parameters of this Contribution Graph (e.g., determining the weight or imprtoance of a contribution) are determined by the project’s community .
3.1. Consent-based Decision-making
3.1.1. Integrative Consent
Integrative Consent decision-making is a new form of consent-based decision-making developed by Round Sky Solutions Cooperative .
For background, consent-based decision-making is a decision-making process where instead of seeking to reach a compromise among members on a proposal (i.e., consensus-based decision-making), we seek to determine if there are any meaningful objections to a proposal (i.e., looking for a good-enough solution). Primarily, instead of trying to reach an agreement that satisfies everyone’s personal preferences, we are looking to find solutions everyone can live with (i.e., the decisions that members can tolerate) .
Concerning decision-making, consent appears the most similar to Quadratic and Collective Voting
The Integrative Consent process is applied in 4 steps:
comments and questions
A member (“proposer”) brings an item forward for the team to consider.
Comments and Questions Step
Members may submit feedback on the proposal
The proposer amends the proposal based on feedback in the Comments and Questions Step that are beneficial to the team.
The members propose meaningful objections to the proposal based on a validation criteria (left to be determined by the members). The objector and proposer will work through how to integrate the objection into the proposal.
When there are no more objections raised by other members, the proposal has passed.
3.2. Community-Market Hybrid Mechanisms
Futarchy is a community-market hybrid decision-making model developed by Robin Hanson, primarily for democratically-governed nations.
In Futarchy, the community utilizes democratic processes to announce desired policies on national welfare, while a market-based mechanism (e.g., betting market) is utilized to estimate whether the desired policies will improve or not improve national welfare .
Futarchy explicitly relies on the assumption that speculative markets are better at aggregating information than democracies .
Organizations built with OrgTech provide a great design space for testing the efficacy of Futarchy as an appropriate means for decision-making .
Level K is currently working on developing a Futarchy model for decentralized autonomous organizations (DAOs) .
3.2.2. Holographic Consensus
Holographic Consensus (HC) is a community-market hybrid decision-making model developed by Matan Field of DAOstack, with contributions from Ezra Weller, Alex Zak, Doug Kent, Adam Levi, and Dean Eigenmann .
Holographic Consensus aims to be a scalable decision-making model for decentralized autonomous organizations (DAOs) that seeks to rely on relative-majority voting systems for deliberation and acceptance of proposals .
In Holographic Consensus, the goal is to curate the attention of the collective onto certain proposals, and utilizing a predictors network (not so dissimilar from Futarchy) to decide on the best proposals . By employing a two-token system, one for attention, and the other for filtering proposals .
A proposer (and any other member) can curate the attention of the collective by staking attention tokens to boost the proposal for consideration .
The cost of a boosted proposal must be higher than the internal value of a proposal (i.e., the total value of promoter approval of a proposal), and the cost is exponential in relation to the number of proposals boosted.
The prediction token is for predictors to stake on the outcome of proposals .
For predictors who staked for good proposals (proposals that passed) or staked against bad proposals (proposals that failed), they will receive compensation .
For predictors who staked for bad proposals or against good proposals, they will lose their stake .
Predictors play 3 roles in decision-making:
Filtering out good proposals;
Signalling and balancing against bad proposals;
Maintain the voting process once staked on the outcome of a proposal .
Predictors are not required to be members of the DAO .
In Part 2 of HC, Matan Field discusses a simple implementation of the HC model in the Genesis V0.2 protocol  [See 10 for greater discussion of technical specifications].
The basic mechanics (interactive smart contracts for assets and rights) of the Genesis V0.2 protocol are:
proposing: submitting a proposal to the DAO by staking tokens (e.g., reputation) via interacting with a smart contract
voting: anyone with reputation (i.e., voting power granted by the DAO) can vote on proposals
predicting: anyone with or without reputation may predict on the outcome of a proposal by staking GEN tokens .
The DAO downstakes (i.e., stakes GEN against the proposal exhibiting a certain amount of disapproval for each proposal) “a specific amount of GEN on each and every proposal to help incentivize initial predictors to search for good proposals” .
The logical flow of decision-making can be summarized as 1) queuing, and 2) resolution .
All proposals start here and require absolute majority of reputation to pass or fail, unless the proposal times out (i.e., passes the deadline) because it met the boosting condition or it does not receive an absolute majority (in either case, predictor’s stakes are returned).
If the boosted condition is met, then the proposal will enter the Boosted Queue. In the Boosted Queue, proposals have a shorter time period for voting, and only require a relative-majority for passage/failure.
A proposal will either be accepted, rejected, or time out
All proposals have Confidence (“how likely predictors believe it to pass”), described in the formula below .
Confidence = Total Approvals / Total Disapprovals
The Boosting Condition is a Confidence score above the minimum threshold for a certain period of time1 .
¹  “The boosting threshold we use in Genesis v0.2, grows exponentially with the number of currently boosted proposals, N_B, with α >1 being the boosting difficulty parameter. The dependency on N_B ensures defensability against the dilution of voters’ attention.”
Thank you for reading this article on new methods for Collective Action that have arisen in Web 3.0.
Part 2 of this article will discuss Organization Technology (OrgTech).
 Handbook for Radical Local Democracy