Electoral Systems Simulator: An Open Framework for Comparing Electoral Mechanisms Across Voter Distribution Scenarios

This paper introduces \texttt{electoral\_sim}, an open-source Python framework that simulates and compares various electoral systems across diverse voter preference distributions using Euclidean distance to the geometric median as a primary performance metric, including a novel theoretical benchmark based on a Boltzmann softmax kernel.

The Gist

Imagine you are trying to organize a massive, chaotic potluck dinner for 5,000 people. Everyone has different tastes: some want spicy food, some want sweet, some want vegetarian, and some want meat. The goal is to pick a menu that makes the most people happy, or at least minimizes how unhappy anyone is.

This paper is about a digital "Potluck Simulator" built by a researcher named Sumit Mukherjee. It's a computer program designed to test different ways of voting to see which method picks the best "menu" (or leader) for different types of crowds.

Here is a breakdown of the paper using simple analogies:

1. The Map of Tastes (The Ideological Space)

Instead of just asking people "Do you like pizza?", the simulator puts everyone on a giant 2D map.

• Left/Right Axis: Think of this as "Budget" (Spending vs. Saving).
• Top/Bottom Axis: Think of this as "Rules" (Strict vs. Free).

Every voter is a dot on this map. Every candidate is also a dot. The closer a candidate is to a voter, the more that voter likes them. The simulator assumes people vote for the candidate closest to them on the map.

2. The Goal: Finding the "Sweet Spot"

The researchers wanted to know: Which voting system picks the candidate closest to the "center of gravity" of the crowd?

They call this the Geometric Median.

• Analogy: Imagine the voters are magnets. The "Geometric Median" is the exact spot where a magnet would feel the least total pull from all the other magnets combined. It's the true center of the crowd, even if the crowd is split into two angry groups on opposite sides.

3. The Contenders (The Voting Systems)

The simulator tested 10 different ways to pick a winner, ranging from the familiar to the weird:

• The "Popular Vote" (Plurality/FPTP): Everyone picks their favorite. The one with the most votes wins.
  • The Flaw: If the crowd is split 40/40/20, the 20% winner takes all, even though 80% didn't want them.
• The "Runoff" (Two-Round & Instant Runoff): If no one gets 50%, the loser is eliminated, and their votes go to the next choice.
• The "Scorecard" (Score & Approval): Instead of picking one, you give everyone a score (1–5 stars) or a "Yes/No."
• The "Math Wizard" (Condorcet/Schulze): This method asks, "If Candidate A fought Candidate B in a one-on-one duel, who would win?" It finds the candidate who beats everyone else in head-to-head matchups.
• The "Proportional" (PR): Instead of picking one winner, it picks a whole team (a legislature) that matches the crowd's makeup.
• The "Magic" System (Fractional Ballot): This is the paper's big experiment. It's a made-up system where your vote isn't a single point; it's a cloud of influence. You give a little bit of your vote to your favorite, a tiny bit to the next best, and so on. It's like spreading peanut butter on toast rather than dropping a single dollop.

4. The Scenarios (The Different Crowds)

The researchers didn't just test one crowd. They simulated 8 different types of societies:

• The "Consensus" Crowd: Everyone agrees on the center (like a small, happy town).
• The "Polarized" Crowd: The crowd is split into two angry camps (like the US or UK today) with a huge empty space in the middle.
• The "Chaotic" Crowd: Many small groups with different ideas (like a fragmented parliament).

5. The Results: What Worked?

The Bad News:
In a Polarized Crowd (two angry camps), the standard "Popular Vote" (Plurality) fails miserably. It picks the candidate from the slightly larger camp, ignoring the fact that the "true center" is actually in the empty space between them. It's like picking the winner of a tug-of-war even though the rope is snapped.

The Good News:
Systems that look at the whole picture (like Condorcet, Score Voting, and Ranked Choice) do much better. They find the candidate who is "least disliked" by everyone.

The "Magic" System (Fractional Ballot):
The made-up "Fractional Ballot" system was the champion. Because it spreads your vote out like a cloud, it naturally gravitated toward the geometric median (the true center) in almost every scenario.

• Why it matters: It acts as a "gold standard." It shows us the theoretical best possible outcome. If a real-world system gets close to this, it's doing a great job.

The One Failure:
The "Magic" system failed in a Dominant Party scenario (where one huge group controls everything). Because the biggest group was far away from the center, the "cloud" of votes got dragged toward them. This teaches us that no system is perfect; if one group is too big and too far from the middle, they will always win, no matter the math.

6. Why This Matters

The author isn't just playing with numbers; they built a toolkit (a Python software package) that anyone can use.

• For Politicians: It helps test if a new voting law will actually work before passing it.
• For Researchers: It's a playground to see how different crowds react to different rules.
• For You: It proves that the way we vote matters a lot. In a divided country, the "first-past-the-post" system might be making things worse, while other systems could find a middle ground.

The Bottom Line

This paper is a simulation lab for democracy. It shows that while no voting system is perfect, some are much better at finding the "center of the crowd" than others. The "Fractional Ballot" is a cool, theoretical idea that shows us what the perfect system would look like, helping us judge how good our current systems really are.

Where to find the tool: The code is free and open on GitHub, so anyone can download it and run their own "potluck" experiments.

Technical Summary

Here is a detailed technical summary of the paper "Electoral Systems Simulator: An Open Framework for Comparing Electoral Mechanisms Across Voter Distribution Scenarios" by Sumit Mukherjee.

1. Problem Statement

The design and comparison of electoral systems in political science and social choice theory suffer from methodological fragmentation. While theoretical results (e.g., Arrow's Impossibility Theorem, Median Voter Theorem) characterize individual systems in isolation, they do not provide comparative rankings across diverse real-world electorate configurations. Conversely, empirical studies are often confounded by strategic voting, endogenous candidate positioning, and the inability to observe counterfactual outcomes. There is a lack of accessible, extensible tooling to systematically simulate and compare a broad range of electoral mechanisms against diverse voter preference distributions using a unified, quantitative metric.

2. Methodology

The paper introduces electoral_sim, an open-source Python framework designed to simulate elections in a two-dimensional ideological space.

A. Spatial Model & Preference Derivation

• Ideological Space: Voters and candidates are represented as points in a unit square $[0, 1]^2$.
  • Dimension 1: Economic Left–Right axis.
  • Dimension 2: Social Libertarian–Authoritarian axis.
• Preference Distance: The distance between voter $i$ and candidate $k$ is the Euclidean distance $d_{ik} = \|v_i - x_k\|_2$.
• Ballot Derivation: The framework assumes sincere voting. Ballot profiles are derived deterministically from the distance matrix:
  • Plurality: Vote for the nearest candidate.
  • Ranking: Candidates ordered by ascending distance.
  • Score: Linear mapping of distance to a score in $[0, 1]$.
  • Approval: Approve all candidates within a fixed distance threshold of the top choice.

B. Evaluation Metric

The primary metric for evaluating an electoral outcome $\hat{x}$ is the Euclidean distance to the Geometric Median ($\mu^*$) of the voter distribution:
$$\delta = \|\hat{x} - \mu^*\|_2$$

• Geometric Median: Computed via the Weiszfeld algorithm. It is chosen over the arithmetic mean because it is robust to outliers and better represents the "center of mass" in polarized distributions.
• PR Systems: For Proportional Representation (PR), the outcome is defined as the position of the median legislator (the candidate at the 50th percentile of the cumulative seat-share distribution). A supplementary metric, the seat-share centroid, is also recorded.
• Secondary Metrics: Majority satisfaction, mean/worst-case voter-outcome distance, and a Gini coefficient for representational inequality.

C. Simulation Scenarios

The framework evaluates systems across eight empirically-grounded scenarios (defined via YAML configuration), ranging from:

• Unimodal Consensus: (e.g., Nordic democracies) where the median and mean coincide.
• Polarized Bimodal: (e.g., Contemporary USA, Brexit-era UK) featuring two distinct voter clusters with a low-density gap.
• Multimodal Fragmented: (e.g., Netherlands, Israel) with many small clusters.
• Dominant Party and Asymmetric Skewed distributions.

D. Systems Evaluated

Ten systems were tested:

1. Winner-Take-All (WTA): Plurality (FPTP), Two-Round Runoff, Instant Runoff Voting (IRV), Borda Count, Approval Voting, Score Voting, Condorcet–Schulze.
2. Proportional Representation (PR): Party-List PR (D'Hondt), Mixed Member Proportional (MMP).
3. Hypothetical Benchmark: Fractional Ballot (a novel mechanism described below).

3. Key Contributions

A. The electoral_sim Framework

The primary contribution is the software architecture itself. It features a modular design where:

• New electoral systems are implemented by subclassing an ElectoralSystem abstract base class.
• New scenarios are defined via YAML files without modifying Python code.
• New metrics can be added to a dataclass.
  This lowers the barrier for researchers to conduct "what-if" analyses on electoral design.

B. The Fractional Ballot (Hypothetical Benchmark)

The paper introduces a novel theoretical mechanism to characterize an approximate upper bound on centroid-seeking performance.

• Mechanism: Each voter's influence is distributed across candidates via a Boltzmann softmax kernel:
  $$w_{ik} = \frac{\exp(-d_{ik}/\sigma)}{\sum_j \exp(-d_{ij}/\sigma)}$$
  where $\sigma$ is a temperature parameter.
• Outcome: The electoral outcome is the weighted centroid of candidate positions based on these fractional weights.
• Variants:
  • Discrete: A single winner is chosen (nearest to the centroid).
  • Continuous: A weighted legislature where power is proportional to the weights.
• Limiting Behavior: As $\sigma \to 0$, it converges to Plurality; as $\sigma \to \infty$, it approaches a uniform mean. Intermediate values approximate the geometric median.

4. Key Results

A. Performance of Standard Systems

• Plurality Failure in Polarization: In the Polarized Bimodal scenario, Plurality performs worst ($\delta = 0.2321$), representing a 53.8-fold increase in error compared to the best system. It consistently elects the candidate of the larger cluster, ignoring the cross-cluster centrist position. IRV and Two-Round Runoff fail similarly in this specific configuration.
• Condorcet/Score/Borda Convergence: In non-polarized (Unimodal, Dominant Party) scenarios, Condorcet, Score, and Borda methods produce nearly identical, superior results to Plurality.
• PR Artifacts: In Multimodal Fragmented scenarios, while the seat-share centroid of PR systems is close to the geometric median, the median legislator position diverges significantly. This highlights a structural limitation where the 50th-percentile legislator in a fragmented parliament does not necessarily represent the voter center.

B. Performance of the Fractional Ballot

• Superior Centroid Seeking: The Fractional Ballot (specifically with $\sigma=0.1$ or $0.3$) consistently achieves the lowest distance to the geometric median across most scenarios.
  • In Polarized Bimodal, it reduced error by 6.5-fold compared to the best standard system.
  • In Multimodal Fragmented, it reduced error by 6.8-fold.
• Failure Mode: The system underperforms in the Dominant Party scenario. Because the largest voter cluster is spatially distant from the geometric median (in the authoritarian quadrant), the centroid-based weighting is "dragged" toward the mass of the dominant coalition, failing to approximate the true geometric median.

C. Stability

Monte Carlo simulations (200 trials per scenario) confirmed that the ranking of systems is stable. Centroid-based methods (including Fractional Ballot) exhibit significantly lower variance in outcomes compared to plurality-based methods, which are highly sensitive to minor fluctuations in cluster sizes.

5. Significance and Implications

• Quantitative Benchmarking: The paper provides a rigorous, quantitative cross-scenario ranking of electoral systems, moving beyond theoretical isolation to practical comparative analysis.
• Validation of Theory: The results empirically confirm theoretical predictions: Plurality fails under polarization, and Condorcet methods excel when preferences are single-peaked.
• New Research Agenda: The Fractional Ballot serves as a theoretical "gold standard" for how well a system could approximate the voter center. Its near-optimal performance suggests that future research should focus on preference-eliciting mechanisms (e.g., using Voting Advice Applications like Stemwijzer) to approximate these spatial distances in real-world settings, rather than relying solely on implicit ballot data.
• Open Science: By releasing the code and framework, the author enables the community to extend the simulation to new systems, higher-dimensional spaces, and more complex strategic voting models.

6. Limitations

• Sincere Voting: The simulation assumes voters always vote honestly. Strategic voting (common in Plurality systems) is not modeled, meaning results represent a "best-case" scenario for each system.
• Dimensionality: The model uses only two dimensions. Real-world electorates involve multiple issue dimensions (healthcare, foreign policy, etc.), which may alter outcomes.
• PR Modeling: The PR outcome metric (median legislator) simplifies the complex reality of coalition bargaining and government formation in fragmented parliaments.

In conclusion, electoral_sim offers a robust, extensible tool for political scientists to analyze electoral design, while the Fractional Ballot case study challenges the community to rethink how voter preferences can be aggregated to better approximate the geometric center of the electorate.