Liquid Democracy (WIP)

This post is a work in progress (WIP) for exploring the use of recommender systems in liquid democracy. Code repository and white paper forthcoming.

Abstract

Liquid Democracy, or delegative democracy, is a form of democracy where voters have the option to delegate their vote to a proxy rather than voting themselves. It combines the merits of direct democracy and representative democracy while limiting the downsides of each. However, voters are still faced with a heavy cognitive load for how to go about exercising their power to delegate decision making. This issue could be addressed through an ecosystem of third party voting advice decentralized applications (“dApps”) where voters instead delegate their decision making to a personalized recommender system model of their choosing. The voter can then review a subset of policy positions taken by their representatives and then rate their performance. The recommender system takes those samples of ratings into account for future recommendations by employing machine learning methods such as matrix factorization and collaborative filtering to optimize for principal-agent relationship. Through the incorporation of differential privacy mechanisms, the liquid democracy platform could allow for political factions and representatives to analyze on a more granular level the shifting policy preferences of their constituents while safeguarding any individual voter’s information. This transparency makes possible a revolution in policymaking preference matching that is unrealizable in today’s system and that could usher in a new standard of democratic accountability.

Links

1. An Introduction to Liquid Democracy
2. A differential privacy framework for matrix factorization recommender systems
3. Can a zero knowledge proof of voting be made using a trusted auth server (without client side crypto)?
4. Surveillance for recommendation system
5. A Differential Privacy Framework for Collaborative Filtering
6. Zero-Knowledge Proof Systems In Anonymous, Publicly Verifiable Boardroom Elections
7. A Fuzzy Recommender System for eElections
8. Voting Advice Applications: Missing Value Estimation Using Matrix Factorization and Collaborative Filtering
9. Evaluating U.S. Electoral Representation with a Joint Statistical Model of Congressional Roll-Calls, Legislative Text, and Voter Registration Data
10. A God of Vengeance and of Reward? Voters and Accountability
11. Decision by Sortition: A Means to Reduce Rent-Seeking
12. Topic-Factorized Ideal Point Estimation Model for Legislative Voting Network