The Oxcart Method

This article expands on the methodology and utilization of representative and direct vote accounting outlined first here and copied herewith at the footer. In this article you will learn 1) my current theory behind the a) legitimacy and b) applicability of this method, 2) the accounting methodology with an example scenario for clarity, and 3) the effects of constraint on the methodology in the form of a) further constraint and b) privacy.

Oxcart Method: Cascading vote-capital accounting for a closed-system of agents fully delegated within that system such that delegations and choice can be changed at any time by individual agents. An efficient accounting-method for legitimizing group-decisions.

This vote-accounting methodology expands on the concept of liquid democracy by extending all delegation-paths; additive vote-accounting is referred to herein as cascade or cascading vote-capital accounting. Requiring everyone in a closed multi-member system to delegate all of their governance tokens to another member (re-callable at any time) prior to a vote ensures that, during the vote, those who have the most accumulated governance capital are the most trusted decision-makers chosen by the community.

Governing and Applied Theory

This section is intended to provide the reader with some basic understanding of what a) legitimizes the use of democratic power and b) how we can use current theory to gain some insight into theoretically applied methodologies.

The Legitimate Use of Group Force and Resources

Legitimacy in a democracy simply refers to the socially approved use of shared force and resources. In a country, force may refer to penalization (e.g. reduced mobility for safety reasons). For most organizations, force refers to constituency (citizen or not). Shared resources are those that are controlled by the constituency or some portion of therein.

Democratic countries use two methods to legitimize the use of force or resources by the few and craft rules around usage for the many: participation and delegation.


Participation in an expressed democratic sense means electing or acting as representative of others; as well as, voting on initiatives directly. The Oxcart Methodology allows any agent to express any choice during an active vote; including revocation of a decision.


This methodology requires agents to delegate their vote to another agent, essentially creating an ongoing elective delegation action of selection, but not a choice abstain from delegation entirely.

Transparent Paths to Legitimacy

Requiring agents to self-select both their trusted representative, and allowing an unfettered path to participation during active votes, would allow the two most powerful methods of communicated democratic choice to be expressed simultaneously. Accounting for the individuals that affirm a course of action and those that support via election of representatives has proven a challenge at scale.

The methodology here allows agents to self-select how they wish to support an endeavor in a manner that is in accordance with both natural law and democratic custom. The methodology requires no outside agent reconfiguration once the agents are known and delegations have been set. In all cases the registered choices and paths to support are both clear and clearly communicated.

Governance as Balance

Agent-interaction alone would determine the level of legitimized representative expression. In a binary system we would see graphs of deciders and supporters on both sides of an issue, yielding something akin to an election of subject-matter-specific representatives, should further debate be warranted. In essence, faster group expression would allow for more robust assessment of the matter by those most trusted in the community. Or in short, one might expect cyclical engagement to have the effect of inspiring greater consensus and engagement toward same, especially among continuously honest and motivated agents in the same system.

Application of Methodology

This accounting methodology would be best suited to already aligned communities, or those looking to develop increased alignment over time. Given that support of initiatives and individuals could be adjusted throughout a vote we would expect to see engagement throughout; given that most or all voting capital would be accounted in the first few votes we would expect to see a push toward consensus develop more meaningfully than in existing accounting methodologies as votes progress.

We would expect to see delegation loops in which no active capital is used, or apathy loops as we’ve termed them. These could be due inattention or through strategic dissent (you delegate to me, I delegate to you, neither registers a vote). Constraint on the system or iterative measures could be used to minimize the likelihood or impact, but otherwise it is state information.

The metrics and graph data are expected to be enviable outputs from this methodology relative to all existing. A dashboard could actively show who the most supported deciders are on both sides of a binary issue. The restive state of a system would show generalized trust-paths; the active state would allow-for and likely inspire robust informed participation by each agent.

Further, the nature of this methodology allows for easy comparative assessment of democratic systems along the lines of liquidity of representation, liquidity of choice, and accounting re vote-transportability and -transparency.

This methodology could inspire some of the first truly self-formed sovereign communities to develop around initiatives OR principles. Existing communities could also use this tool to garner valuable feedback to move toward more equitable governance options otherwise. Ultimately we expect to see AI, ML, and other non-human feeds enter into considered (designed) delegation chains to help inform better group outcomes.

Iteratively we expect step-changes in the speed and efficacy of DAOs requiring human decisions and other peer-controlled constructs requiring human feedback. Automated systems that use this methodology could theoretically produce meaningful output on each decision very quickly (on the order of minutes) allowing for advancement of process-oriented initiatives; with the addition of automated incentive we could see the first DAOs to self-govern by auto-incentivizing majority alignment directly, or automated-forking along constituency-lines for contentious votes.

Liquid Proposal Power

An open automated system of self-governance would require that all agents be able to change the methodology of reconstitution or execution, and so would require forms liquid proposal power as well, not discussed explicitly here.

Detailed Methodology

Terminology Fig. 1a
Terminology Fig. 1a

The Methodology outlined here treats a user’s voting capital (ref: ‘voting weight’) as a fully transferrable token of one’s support of a stated endeavor or reconstitution of members. The holder of this token must elect to place it in support of a delegate (also: proxy, representative) by default to be considered constituent themselves, and can otherwise elect to use it along with any agglutinative voting capital that has been delegated to them. Both of these actions are unrestricted in that a user can change their delegate at any time (fully liquid representation) or their decision (fully liquid democracy) in a matter being actively decided.

Terminology Fig. 1b
Terminology Fig. 1b

The accounting of tokenomic votes in this methodology are such that no votes may be unutilized; unused DVC carries fully to the chosen delegates in a cascading manner such that any agent executing a vote has an AVT value equal to their total CVC + their DVT. There are cases where not all DVC becomes AVC and those will be discussed below; in a fully-connected system all CVC becomes AVC upon execution of the first vote.

Example Fig. 2a
Example Fig. 2a

The straight arrows in Fig 2a represent the delegations chosen by agents such that Agent 1 is delegating to Agent 2, and so on with Agent 6 delegating to Agent 1. This would be the INACTIVE STATE of the mechanism resulting from this methodology in that all agents are delegated (considered constituent) but no active proposals require decision, or none has been registered.

Example Fig. 2b
Example Fig. 2b

The curved arrows in Fig. 2b,c,d represent the choices of the first and only deciders in this example; specific choice is not shown, just that a decision was registered by an agent. In Fig 2b Agent 2 registers a decision and this accounts for 100% of the available voteable capital in the system given that it’s a closed system.

Example Fig. 2c
Example Fig. 2c

Agent 6 registering a decision (Fig. 2c) would then mean that the available delegations from Agents 3, 4 & 5 would then be utilized along their delegation path by Agent 6, and Agent 2 would lose the delegated voting capital of Agents 3 through 6.

Example Fig. 2f
Example Fig. 2f

In the final action of this example (Fig. 2f), Agent 4 changed their delegation from Agent 5 to Agent 2 (motivations here could be many, but result here in choosing the new representative). We can see that this changes this again places Agent 2 in control of the most active voting capital. Agent 2 would dictate the outcome of this vote should a standard 50%+1 threshold be applied. This final stage also highlights the benefit of some constraints in applied systems discussed further below.

Further Constraint

Constraining the methodology applied above in manners apart from what we’ll call the ‘ideal form’ (fully liquid, transparent, and cascading without modifiers; term coined by Steven Vitka ref. below) provides some insight into how parameterization could greatly impact outputs of a constrained system, or be used to model existing democratic mechanisms.

Timing - Limiting either choice or delegation during a portion of the voting-period could be useful; perhaps freezing delegation for a time before vote-end.

Constrained Delegation - Limit the delegates that can be chosen; such as a departmental approval.

Forced Representation - Don’t allow some agents to decide directly (delegate only).

Delegation Incentives or Penalties - Add or subtract delegated voting capital. This could be dependent on the number of ‘hops’ (modifying the DVC according to order in the delegation chain). For example a decay-factor could be applied to account for sentiment akin to: “I trust you, but I don’t trust your choice of friends.”

Fungible delegations - Allowing an agent to split their delegated voting capital between multiple other agents. This could be useful in lower-trust environments.

Automated Constitution and Execution - This methodology would be well-suited for blockchain-based accounting methods given that human voting mechanisms require a high degree of trust and use across broad geographical regions. This method would additionally allow for incentive to be applied directly to actions considered valuable to the group that can be measured directly or in concert with other available information. I believe this methodology could be used to fully automate a DAO dependent on strategic choices and agreement among members; including agent entry and exit (onboarding/off-boarding).

Automated Delegation - Designing for automated delegates within the system might prove useful, such as state-dependent mechanisms that register a decision under conditions. Human Agents with the ability to recall such DVC would help abate some current concerns relating to the artificial control of digital and physical human environments.

Automated Proposition - Static or dynamic resourcing could be applied in many ways to supported individuals or cohorts.

Privacy - There are several privacy implementations that will impact agent activity such as shielding all constituency activity from outside view (akin to a black box) versus shielding agent-level interactions from each-other (such as: delegation state, decision state, outcome state).


When evaluating the legitimacy of democratic systems, three key dimensions to consider are: 1) vote portability and accountability, 2) available choices and actions, and 3) data availability and trust in representatives. Unfortunately, most physical and digital decision-making systems fall short in one or more of these dimensions to deleterious effect on Agent engagement. However, by using this comparative framework, we can have more nuanced discussions about the strengths and weaknesses of different applied systems.

While nothing is certain without testing, intuition dictates that cultural and mathematical properties align in such a way to allow for further exploration of the mechanism and parameterization. Considering the fully liquid and cascading version the ideal form of the methodology, it could allow for self-legitimized agent-level formalization of appendant parameters with high degrees of speed and flexibility.

Proposal power is only discussed broadly here. Being able to surface a proposal using the same mechanism would essentially make the mechanism relatively Turing Complete, efficient, and culturally legitimizing. While the use-case has not nearly been proven, I believe this methodology seeds the first fully on-chain DAOs with encoded proposal-, decision-, and execution- ability.

Gratitude & Resources

I have received no direct funding for these efforts to date. I am an Architect and Engineer by education, practiced in infrastructure and high-rise construction financial monitoring and risk controls. I contributed to The Index Cooperative DAO toward contributor connection, strategy, and governance operations. I plan to continue working as a DAO Mechanism Designer, a craft I enjoy. I’m grateful to be working with so many great minds of our day toward more perfect communication.

That said, the terms we use for various organizational and decision-making mechanisms vary widely and do not serve us well. The design of systems of constituted voting rights require extensive testing and research. I hope to see the academic and web3 communities continue to convene more regularly to that end. The application of privacy settings, parameterization, and automation will greatly impact the usage and outcome of any system based on this methodology.

Join the discussion here.

I’d like to thank the following for their inspiration, passion, experience, and ongoing encouragement around this endeavor. I was trying to solve a big problem for one DAO and found something much larger in the seeking. I can’t possibly account for all positive influences here, but I hope to cover the most direct:

frogmonkee (Bankless DAO): For the initial inspiration at ETH Denver ‘22.

Manon Revel (MIT): For surfacing outcome efficacy of in-process delegation.

noturhandle.eth (Butter): For thoughtful high-context engagement on the design.


COORDINATIVE and DAO Governance Collective; especially Feems and coin.

Index Coop DAO; especially Shawn Grubb and Meg Lister.; especially Nathan Schneider, Michael Zargham, Jessica Zartler, & Joshua Tan and Steven Vitka for surfacing his Patent Application and deeply discussing the ideal form of his design.

Token Engineering Academy; especially octopus.

Original Post

Force Constrained Infinitely Liquid Delegation Mechanism
Force Constrained Infinitely Liquid Delegation Mechanism


Feedback welcome. Published cc0 on 2023.03.15 by mel.eth. Parked for self-education purposes. Not advice in any context.

Subscribe to mel.eth
Receive the latest updates directly to your inbox.
Mint this entry as an NFT to add it to your collection.
This entry has been permanently stored onchain and signed by its creator.