Quantifying Gerrymandering With Simulated Annealing

TL;DR

This paper introduces a novel method combining Markov Chain Monte Carlo with Simulated Annealing to quantify gerrymandering, demonstrating its effectiveness in analyzing district fairness efficiently.

Contribution

It presents a new approach using Simulated Annealing within Markov Chain Monte Carlo to measure districting fairness, improving speed and accuracy over previous methods.

Findings

Successfully applied to Texas redistricting data

Demonstrated rapid generation of representative voting distributions

Indicated bias in enacted districting plans

Abstract

Gerrymandering is the perversion of an election based on manipulation of voting district boundaries, and has been a historically important yet difficult task to analytically prove. We propose a Markov Chain Monte Carlo with Simulated Annealing as a solution for measuring the extent to which a districting plan is unfair. We put forth promising results in the successful application of redistricting chains for the state of Texas, using an implementation of a redistricting Markov Chain with Simulated Annealing to produce accelerated results. This demonstrates strong evidence that Simulated Annealing is effective in quickly generating representative voting distributions for large elections, and furthermore capable of indicating unfair bias in enacted districting plans.