How Simple Choices and Noise Make Compactness Tests Unreliable for Detecting Gerrymanders

Political district scores are meant to flag unfair maps, but numeric measures respond not just to map shape — they also respond to noise and to seemingly innocuous implementation choices. Measurement values are affected by both noise and the compounding effects of seemingly innocuous implementation decisions. Such issues will arise for any measure, complicating efforts to use a single metric as a legal or policy standard.

🧭 What Was Studied:

A case study focuses on commonly used geometric compactness measures for district boundaries. These measures are intended to provide concrete axes for comparing districts and detecting gerrymanders, but their outputs can be altered by factors that are irrelevant to fairness or compliance with civil rights law.

🧪 How the Analysis Worked:

• Demonstrations show how small sources of noise and specific implementation decisions change compactness scores.
• Scenarios illustrate how an adversary could manipulate measurement choices to affect the assessment of a given plan.
• This paper accompanies the release of packages in C++, Python, and R that correctly, efficiently, and reproducibly calculate a variety of compactness scores.

🔎 Key Findings:

• Compactness metrics produce concrete comparators but are sensitive to noise and implementation details.
• Several widely used geometric compactness measures are affected by factors unrelated to map fairness or civil-rights compliance.
• An adversary can exploit implementation flexibility to change measured outcomes and thereby influence evaluations of a plan.
• The instability in scores undermines reliance on a single compactness measure as a definitive legislative or judicial standard.

⚖️ Why It Matters:

The fragility of compactness scores suggests caution when adopting them as legal or policy thresholds to police redistricting. Courts and legislatures should recognize that numeric measures can be altered by benign-seeming choices or by bad actors, and that reproducible, transparent implementations (such as the provided C++, Python, and R packages) are essential for fair and consistent use of compactness metrics.