Simulating the Quality of Community Decisions

How Our Project Will Contribute To The Growth Of The Decentralized AI Platform

Photrek will deliver a high-quality governance simulator to assure the long-term integrity of the Deep Funding program. Situational awareness regarding how proposed improvements to the voting, correlation detection, and reputation regards will allow designs to be tested prior to implementation. The Photrek team are international leaders in the simulation of complex community sentiment which can be used to drive understanding of the characteristics of different voting systems.

Company Name (if applicable)

Photrek

The core problem we are aiming to solve

Public treasury decision-making is an exceptionally difficult problem. Each member of a community needs to be able to express their preferences in a manner that contributes to maximizing community value. A variety of problems arise from poorly designed voting systems, including:

1. splitting of communities into majority and minority factions;

2. diversion of public resources toward private gains;

3. funding of low impact projects.

SingularityNET has made a commitment to address these problems through the use of quadratic voting, graded ranking of proposals, and reputation scores; however, the current implementation does not match evidenced-based designs and thus could be improved.

Our specific solution to this problem

Photrek proposes the development of Monte Carlo simulation of agents that model the actions and interactions of multiple individuals forming opinions within a complex social community. This computational simulation foundation will be used to design and analyze voting methods for decentralized communities. The critical aspect of the complex network approach relies on the patterns of connection between their elements that are neither purely regular nor purely random, as the real-world networks of people are structured [9-10].

Photrek will accomplish the following objectives.

1. Establish a foundation for simulating the interaction between community members that establishes clusters of sentiment regarding opinions about proposals.

2. Simulate community members with a distribution of sentiments and AGIX holdings voting positive, negative, or abstain on sets of 2 to 10 proposals.

3. Utilize the complex network of sentiments to simulate voters expressing their strength of opinion on projects with and without quadratic costs on the strength of opinion.

4. Develop a proposal plan for how correlation measurements can be incorporated into the simulation for collusion mitigation.

5. Develop a roadmap that includes incorporation of reputation scores into the simulation capability.

In order to assure the simulation results have high integrity and community alignment, Photrek will set aside $5000 for a red team review of the results.

Project details

Objective

Photrek will develop a Monte Carlo simulation [1-2] capability to determine the ability of voting systems to maximize the quality of community decisions. Photrek will compare two ‘strength of conviction’ methods for ranking the quality of proposals. Both methods will distribute voting credits based on the square-root of AGIX holdings. Quadratic or Plural Voting [3-4] requires the voter to rank their opinion on each proposal by distributing their credits. The votes applied to a proposal is the square-root of the credits applied. “No-cost Grading” applies the voter’s credits equally to all proposals and requires the voter to separately grade the proposal on a scale of 1-10. 

We seek to evaluate the claims of Plural Voting that the application of the quadratic cost to preferences incentivizes individual voters to express the truth about their convictions rather than strategically using extreme opinions. A second claim that will be evaluated is whether Plural Voting mitigates the formation of dominant factions across a series of decisions. The goal is to encourage pluralistic values within decentralized communities by assuring that both minorities with strong convictions and majorities with broad support can influence outcomes without a particular faction gaining permanent control.

Approach

Andre Vilela and a graduate student will develop statistical agent-based models to simulate community sentiments [5] about a group of proposals. The proposals will have a distribution of value (the quality returned to the community) and cost (the funding requested). Vilela utilizes majority-vote dynamics to model how gossip within a community leads to dynamic variation in community sentiment [6-10]. The Monte Carlo simulation foundation will be applied to Plural Voting and No-Cost Grading for the purpose of determining the ability of these voting methods to maximize the expected community profit (value - cost). A roadmap incorporating reputation will be provided. Vilela has successfully applied these methods to a variety of socioeconomic systems including social influence, network structures, information diffusion, and consensus dynamic. Photrek with Vilela completed contracts with IOTA on blockchain consensus and Cardano Catalyst on diversified voting [11-12]. 

Nelson and Attieh will work with the SingularityNET community to apply simulation results to a requirements roadmap. To assure an independent critique of the results, Photrek will provide $5000 for 2-3 experts selected from the SingularityNET community to provide a technical review.  Starting with our second report, the reviewers will provide an independent critique and recommendation in each report.

References

[1] Short-time Monte Carlo simulation of the majority-vote model on cubic lattices. K. P. do Nascimento, L. C. de Souza, A. J. F. de Souza, André L. M. Vilela, H. Eugene Stanley. Physica A - Statistical Mechanics and its Applications, 2021.

[2] Three-state Majority-vote Model on Small-world Networks. Bernardo J. Zubillaga, André L. M. Vilela, Minggang Wang, Ruijin Du, Gaogao Dong, H. Eugene Stanley. Scientific Reports, 2022.

[3] Quadratic Voting. Steven P. Lalley and E. Glen Weyl. Available at SSRN, 2014.

[4] Posner, Eric A, and E Glen Weyl. 2015. “Voting Squared: Quadratic Voting in Democratic Politics.” Vanderbilt Law Review 68 (2). 

[5] Effect of Strong Opinions on the Dynamics of the Majority-Vote Model. André L. M. Vilela and H. Eugene Stanley. Scientific Reports, 2018.

[6] Opinion dynamics in financial markets via random networks. Mateus F. B. Granha, André L. M. Vilela, Chao Wang, Kenric P. Nelson and H. Eugene Stanley. PNAS, 2022.

[7] A Three-state Opinion Formation Model for Financial Markets. Bernardo J. Zubillaga, André L. M. Vilela, Chao Wang, Kenric P. Nelson, H. Eugene Stanley. Physica A - Statistical Mechanics and its Applications, 2021.

[8] Majority-vote model with limited visibility: An investigation into filter bubbles. André L.M. Vilela, Luiz Felipe C. Pereira, Laércio Dias, H. Eugene Stanley, Luciano R. da Silva, Physica A - Statistical Mechanics and its Applications, 2021.

[9] Three-state Majority-Vote Model on Barabási-Albert and Cubic Networks and the Unitary Relation for Critical Exponents. André L. M. Vilela, Bernardo J. Zubillaga, Chao Wang, Minggang Wang, Ruijin Du, H. Eugene Stanley. Scientific Reports, 2020.

[10] Majority-vote model for financial markets. André L. M. Vilela, Chao Wang, Kenric P. Nelson, H. Eugene Stanley. Physica A - Statistical Mechanics and its Applications, 2019.

[11]  IOTA Majority Vote Final Report. André L.M. Vilela and Kenric P. Nelson, 2020. https://docs.google.com/presentation/d/1Iu47WIhV2AX7uP0pJxG3uW1G0R8vmggkz-l3vFz8OQM. Accessed February 4, 2024.

[12] Diversify Voting Influence Close Out Report. Photrek, LLC. Kenric P. Nelson, André L.M. Vilela, and Megan Hess, 2021. https://docs.google.com/document/d/1ibAZBDgj3cQGEcAePQ_Qjl5wJvMICLqGEU65CyskDr4/edit?usp=drive_web&ouid=106499308022823206372&usp=embed_facebook.

Open Source Licensing

Additional links

https://photrek.io

https://andrevilela.com/

https://www.dropbox.com/s/922kdcb0facv8kw/ScreenFlow00.mov.zip?dl=0

Additional videos

Presentations of Cardano Catalyst Diversify Voting Infuence:

DVI Video Final Report - 5 minutes

DVI Video Final Report - 50 minutes