The Politics of Romance

What is the Love Machine?

Last year I began an experiment in algorithmic dating, where I matched students at Georgia Tech with a date using machine learning. The algorithm takes a survey of student preferences and uses it to find the perfect pairing. You can read more about the experiment here.

Round 2 of the Love Machine went well this year — we drew about 2,000 submissions from the student body (out of 32,000 students total). I invited 200 participants to attend a Valentine’s Day dinner where they could meet their match. Only 22 participants showed up, so I did not have enough data to analyze the accuracy of the pairings. Despite not having feedback from the matches, I have abundant data on student demographics and preferences, which I have been exploring for the past several weeks.

So, what did you decide to do with the data?

Given the imminent presidential election, I thought it would be fun to take a look at how politics play into dating at Georgia Tech. I know many people who refuse to date someone of a different political affiliation and I had assumed this was the consensus among my age group.

After reading the comments on my Reddit post concerning the Love Machine, I began to question my assumptions — perhaps the story is more complex. In this response, we have two arguments. One emphasizing that political values are not an important aspect of a relationship, the other pointing out that perhaps we shouldn’t turn a blind eye to how politics affects our partner’s choices.

This led me to the question driving my research, is politically biased dating generalizable to the greater student population, or is it a quirk of my social circle. With this objective in mind, I began to comb through my data. First, I visualized some demographic info, so I could get a feel for the political atmosphere at Tech.

Interesting — living in Georgia I would expect more Republicans. I wonder how this breaks down by major:

I showed this chart to my roommate. She thought it was interesting but too detailed. What she really wanted to know was whether or not Business was the most Republican major:

Wow, she was right.

Now that we’ve seen the political breakdown of our campus, let’s move onto my main question about the data:

Does politics impact dating at Georgia Tech?

Okay, the answer to that is yes then …2/3 of students have political biases when it comes to dating. Now for the juicier question, who dislikes who?

Check this out:

This chart shows inter-group tensions: Republicans tend to not want to date socialists, anarchists, and communists and vice versa. However, on the whole, Republicans are a much more politically biased group. And on the whole, socialists, anarchists, and especially communists tend to be discriminated against.

Ok, so which group is the pickiest when dating? And which group is the least desired?

People are biased, and Republicans are the most biased of them all. I wonder whether these biases are built on actual differences in values, or whether we are a lot more similar than we think.

Looking over this visualization, I had a lot of comments, so I took the liberty of drawing all over it to point out some observations:

• What surprised me was just how similar the Republicans and Communists were. They had more consensus than disagreement despite eliminating each other from their pools.
• On average, the communists tended to care more about money than almost any other political affiliation, they also believed strongly in gender roles. These values oppose the elimination of class injustices and gender inequality I commonly associate with communism.

While these observations were surprising, it is important to keep in mind the sample sizes for some political affiliations were small (communist = 9, anarchist = 11, green = 13).

What can we conclude?

These charts tell me the political polarization so apparent in U.S. discourse today, is not confined to select portions of our lives — how we consume our news or who we vote for, but political bias seeps into our social lives and impacts who we form deep connections with. Students (including myself) were willing to discard an entire pool of possibilities based on preconceived notions of a political group.

In the episode of NPR’s Invisibilia, “Bubble-Hopping”, a Google engineer named Max ponders deep into the night about the way he socializes and how that drives the way he ultimately lives. He deems himself, and others in American society, as trapped in a “preference prison”, one that only provides a set of experiences we prefer but not one representative of all the possible situations we can experience.

Max, like me, enjoys creating algorithms to solve his problems, so he created an app to scrape Facebook and send him to random events, letting no biases get in the way, and he has been living his life happily random ever since.

After listening to this program, I felt a vague sense of guilt about my algorithm. All this time, I believed I was creating a bit of randomness in the GT dating pool, when I was perpetuating patterned behavior, reinforcing the bubbles Max has been trying so hard to pop.

I will graduate Georgia Tech as of May 2020, so I will be passing off the Love Machine project to new hands. It wouldn’t be a proper goodbye without some advice, so I will leave you with this:

When you create next year’s survey, think of ways to help people break out of their bubbles. Perhaps they haven’t found their “soulmate” yet because their bubbles are too small and they need some helping in expanding them.