Ideological Tolerance, Voter Abstention, and Party Polarization: An Agent-Based Extension of the Hotelling Model

Imagine a world where politics is like a giant game of musical chairs, but instead of chairs, the seats are ideas, and the players are political candidates. For decades, a famous idea in political science, known as the Hotelling model, suggested that in a two-party race, both candidates would eventually squeeze right into the middle of the room. The logic was simple: to win, you need the most votes, and the most votes are usually found in the center. So, everyone moves to the middle, and the game becomes a boring tie. But in real life, we often see the opposite. Candidates scream from opposite ends of the spectrum, and the middle feels empty. Why don't they just meet in the middle like the old theory says?

This paper dives into that mystery by looking at the people sitting in the audience: the voters. The authors introduce a new twist to the game. In the old rules, everyone had to vote, no matter how much they disliked the candidates. But in this new version, voters have a "patience meter." If a candidate is too far away from what they believe, they simply walk out of the room and don't vote at all. The researchers used computer simulations—essentially creating a virtual world with 2,000 fake voters and two competing candidates—to see how this "I'm not voting if you're too extreme" rule changes the game. They found that when voters are picky and only show up for candidates very close to their own views, the candidates stop meeting in the middle. Instead, they run to opposite corners to grab the specific groups of people who will actually show up, creating a polarized, divided political landscape.

The Game of Political Musical Chairs

Let's set the scene. Imagine a long, straight line representing the entire political spectrum, stretching from "Super Left" to "Super Right." In the classic version of this game, invented by economists and political scientists long ago, voters are spread out evenly along this line. The rule is simple: you vote for the candidate closest to you. Under these conditions, the smartest move for both candidates is to stand right in the middle. If one candidate moves to the left, they lose the right-wing voters but don't gain enough new ones to make up for it. So, they both huddle in the center, and the election is a dead heat.

But the real world is messier. People aren't robots who vote no matter what. Sometimes, if the choices are terrible, people just stay home. This paper asks: What happens if we add that human element? What if voters have a "tolerance threshold"? Think of this threshold as a personal bubble. If a candidate steps outside your bubble, you don't just vote for the "less bad" option; you refuse to participate. You go home, watch TV, and let the candidates fight it out without you.

The researchers built a computer model to test this. They created a virtual election with 2,000 voters. These voters weren't spread out evenly; instead, they were clustered around the middle, following a bell-curve shape (a "truncated normal distribution"). The key variable was how "spread out" these voters were. Sometimes they were tightly packed in the center (low dispersion), and sometimes they were scattered far and wide (high dispersion).

Then, they introduced the tolerance threshold, which they called $r$. This number represents how far a voter is willing to stretch their beliefs to support a candidate.

• If $r$ is small (like 0.05), voters are extremely picky. They only vote if a candidate is practically standing on their doorstep.
• If $r$ is large (like 0.25), voters are more flexible. They are willing to vote for someone who is a bit further away.

The candidates in the simulation were smart agents. They didn't just pick a spot and stay there. They kept testing small moves left or right (a step size of $\epsilon$) to see if they could get more votes. If moving left got them more people, they moved left. If moving right helped, they moved right. They kept doing this for up to 300 steps until they found a stable spot.

The Twist: When Picky Voters Push Candidates Apart

The results of these simulations were fascinating and showed that the old "meet in the middle" rule isn't always true. It depends entirely on how picky the voters are and how spread out they are.

1. The Pickiness Trap
When the tolerance threshold $r$ was small (meaning voters were very strict), the candidates didn't converge. Instead, they diverged. Why? Because if you are a candidate and you know that voters in the middle will only show up if you are very close to them, you can't just stand in the middle and hope for the best. You have to go find a specific group of people who are willing to vote for you. So, one candidate might move left to capture the "strict left" voters, and the other moves right to capture the "strict right" voters. They stop fighting for the center and start fighting for their own little cliques. This leads to polarization, where the two parties are far apart.

2. The Flexibility Factor
However, when the tolerance threshold $r$ was large (voters were more forgiving), the old rule kicked back in. The candidates moved back toward the center. Since voters were willing to support candidates even if they weren't perfect matches, the candidates realized the best strategy was to grab the biggest chunk of the crowd: the people in the middle.

3. The Surprising "U-Shape"
Here is where it gets really interesting. The relationship between how spread out the voters are (measured by standard deviation $\sigma$) and how polarized the candidates become isn't a straight line. It's a curve.

The authors found that as they increased the spread of the voters (making $\sigma$ go from 0.03 up to 0.25), the distance between the candidates didn't just get bigger or smaller in a simple way.

• First, as the voters spread out a little, the candidates actually moved closer together.
• But then, as the voters spread out even more, the candidates started moving apart again.

This "non-monotonic" pattern suggests that the game changes rules depending on the crowd. When the crowd is moderately spread out, candidates might focus on the center. But when the crowd is very scattered, the candidates abandon the center entirely to chase the scattered groups on the edges.

What About Turnout?

The simulation also tracked how many people actually voted. Naturally, when the tolerance threshold $r$ was small, fewer people voted. If you are picky, you stay home. But there was a weird twist here too. The total number of voters didn't just go down as the crowd got more spread out. It went down, hit a low point around $\sigma = 0.15$, and then went back up.

This suggests that in a very divided society, the type of voter who shows up changes. Sometimes, the "marginal" voters (those on the edges) drive the turnout. Other times, it's the "median" voters (those in the middle). The balance shifts depending on how tolerant the voters are and how scattered the population is.

The Bottom Line

This paper doesn't claim to have solved the mystery of politics forever. It's a simulation, a computer experiment that suggests how things might work. But it offers a powerful new way to look at why our political world feels so divided.

The main takeaway is that voter abstention (people staying home) isn't just a side effect; it's a driver. When voters say, "I won't vote unless you are exactly like me," they force politicians to stop trying to be everyone's friend and start trying to be someone's best friend. This pushes the candidates to the extremes, creating the polarization we see today.

The authors suggest that if we want to understand why candidates aren't meeting in the middle, we have to stop assuming everyone votes. We have to realize that the fear of losing the picky voters might be pushing politicians apart, while a more forgiving electorate might pull them back together. It's a reminder that in the game of politics, the people who don't show up might be the ones shaping the game the most.