SwatDeck, Diversity, and the Science of Networks

swatdeck_mainIn 2013, many colleges throughout the country were ripe with student activism. There were fervent demonstrations and strategic protests about sexual assault, climate change, and diversity and intolerance. During my senior year at Swarthmore College, I watched many of these movements unfold. So often, there was a lack of agreement among students, not just on opinions and experiences, but even in how to engage in constructive debate. Campus communities suddenly seemed more siloed than usual, something we have all likely witnessed among friends, colleagues, or family members. Small, subtle differences start to evolve into larger differences, into social walls that are difficult to break down.

Swarthmore’s administration recognized this trend and offered a new grant to fund projects aimed at promoting sustained and meaningful interactions among community members, while cultivating a more inclusive learning environment. That fall, three friends—Isaac Opoku, Emma Kates-Shaw, Raven Bennett—and I had an idea.

SwatDeck founders, clockwise from top-left: Isaac Opoku, Raven Bennett, Brennan Klein, Emma Kates-Shaw

SwatDeck founders, clockwise from top-left: Isaac Opoku, Raven Bennett, Brennan Klein, Emma Kates-Shaw

We noticed that the goals of this grant seemed to tiptoe around a deeper problem with the social networks on campus—not the online kind, but the interwoven cliques, clusters, and friend groups that we witnessed becoming more and more isolated from one another. The networks on campus, the patchworks of friendships, the social fabric, any weaving analogy we could muster, seemed to be unraveling.

Our submission, an attempt to rewire social connections on campus, ended up winning the grant. We called it SwatDeck, from Swat—an abbreviation for Swarthmore College—and Deck—a reference to a deck of cards, as it reminded us how the relatively small number of cards in a deck can be arranged into virtually infinite new combinations.

The goal of SwatDeck was to reshuffle the social networks of Swarthmore by bringing together four random students for an adventure in Philadelphia. On the Friday before their adventure, participating students checked their mailboxes and found a playing card from a standard deck. The next day, card in hand, they met at the Swarthmore train station and found the other three students with their same card to make four of a kind (e.g. all the 9s formed a group of four). Each student was then given $25, a train ticket, and a small notebook. After the expected amount of awkward small talk, the quartets boarded the train from Swarthmore to Philadelphia. Then, the SwatDeck experience began.

With SwatDeck, we wanted to build an easy way for students to learn from and listen to one another. We wanted to inspire new perspectives through unexpected conversations. In the end, we wanted to transmit a mindset that probably checks all the boxes of Millennial Cliché Bingo: embrace serendipity, meet new people, carpe that diem. This is a mindset that is timeless, repurposed here for a new generation of people who are growing to understand that their peers have lives that are just as complex and interesting as their own.

Students reported feeling happier after SwatDeck, more connected to the average Swarthmore student, and a greater sense of belonging on campus.

Early on as we were brainstorming for SwatDeck, we found ourselves inspired by an entire field of science that explores the structure of groups of people, how social networks form, and how information is shared. Using this Network Science research, we began to rethink a problem on campus that, until then, hadn’t been looked at through a networks lens. This research took us back to Königsberg in the 1700s and to a meth-addicted mathematician in the 1940s, but we eventually found our entry into social networks through a psychologist in the 1960s.

It wasn’t always obvious to describe our friends and acquaintances as networks of interconnected relationships. Until the 1960s, there hadn’t been clear scientific ways to study the structures of relationships. But this changed when Stanley Milgram (yes, electric shock Stanley Milgram) explored the “Small World Problem” in his 1967 article in Psychology Today. For this study, Milgram recruited people living in Omaha, Nebraska, and Wichita, Kansas, and gave each of them a package that needed to be mailed to a stranger in Boston, Massachusetts. Since no one knew this mystery person, they needed to send it to someone whose friend (or friend’s friend’s friend) might know them.

Read Stanley Milgram’s 1967 article “The Small-World Problem”

Let’s say you live in Des Moines, Iowa, and you receive a package from your friend in Omaha. In the package, there is a letter telling you to send the package to a friend of yours who might be able to deliver it to somebody in Boston. You know that your friend from college lives in Columbus, Ohio, and she has family that live outside of Boston. You send it to her, she sends it to her uncle, and so on. Milgram and his colleagues found that packages tended to pass through the hands of fewer than six people before reaching its final destination in Boston. Famously, this finding became known as Six Degrees of Separation—the idea that you are connected to every person in the world through at most six social jumps. I say with statistical certainty that, at most, your friend’s friend’s friend’s friend’s friend is Beyoncé, although with greater interconnectivity and our online social networks, the “six” degrees of separation is really more like three.

Aside from being a fun social experiment, Milgram’s research was among the earliest to identify scientific properties of the social networks we form. Importantly, words like friend and relationship became mathematical concepts, measures waiting to be analyzed. In his 1967 article Milgram writes:

“The importance of the [small-world] problem…[lies] in the fact that it brings under discussion a certain mathematical structure in society, a structure that often plays a part, whether recognized or not, in many discussions of history, sociology, and other disciplines…Important historians make the point that in the Dark Ages, communication broke down between the cities of western Europe. They became isolated and simply did not have contact with each other. The network of acquaintances of individuals became constricted. The disintegration of society was expressed in the growing isolation of communities, and the infrequent contact with those living outside a person’s immediate place of residence.”

Fast-forward 50 years, and the growth of computational power along with insights from big data analytics allows researchers to show that the structures of our interconnected webs of relationships have many surprising and profound effects on our lives.

The backbone of SwatDeck lies in the observation that two systems with exactly the same parts but organized differently can display incredibly different properties. Just as a thousand different Lego pieces can combine to form countless unique spaceships and pirate ships, a network of a thousand people organized in new ways can exhibit very different behaviors as a whole. These insights are part of what is now the newly established field of Network Science.

The growth of Network Science was jump-started in the 1990s with the insight that many biological networks and man-made networks often share remarkably similar structures. One key property shared by both man-made and biological networks is homophily, which in simplest terms is the idea that “birds of a feather flock together.” We like people who are like us. As new nodes—the term for the units of a network (e.g. individuals, proteins, websites, etc.)—join the network, they are likely to form links with nodes that are highly-connected, as well as to nodes with similar characteristics. A good example of this is our beloved worldwide web. While most websites link to hubs like Google, think of that Tumblr with pictures of kittens in aprons you love so much—it mostly links to other cat Tumblrs or cooking Tumblrs, which link to other feline adventure websites, and so on, until you zoom out to see a picture that looks similar in structure to this map of the internet from 2003.

“In the Dark Ages, communication broke down between the cities of western Europe . . . The disintegration of society was expressed in the growing isolation of communities, and the infrequent contact with those living outside a person’s immediate place of residence.” -Stanley Milgram

For an offline example, imagine walking into a party where you know no one. Or perhaps more apt, imagine being a freshman in college. Your first friend may be your roommate, then your hallmates, then dormmates. However, as you go through your first year you become friends with students who, like you, played tennis in high school or are in the orchestra or are politically active. You find yourself befriending people who remind you a lot of yourself. This type of attachment is homophily at work. As this college social network grows, new nodes tend to connect to similar nodes or other highly-connected nodes. This structure is found time and time again in networks throughout nature and society, and by identifying and understanding this structure, Network Scientists examine how things like ideas, disease, gossip, and happiness spreads through our social networks.

In a way, homophily moves a network towards a state of order, simplifying complex and random networks into similarly connected clusters. But what happens when this clustering process is allowed to run its course? What happens when a network grows into too much order?

Consider the network of congressional voting trends over the past 25 years, shown below. Each node in this network is a senator, and there is a link between two nodes whenever two senators voted the same way during a session. Over time, the effect is clear—even if the reasons behind it may not be. As liberal senators began voting only with fellow liberals and conservative senators began voting only with other conservatives, two distinct voting networks formed. In a realm where cooperation is vital to accomplish even the simplest of tasks, unchecked homophily can create distinct, often unseen, factions. As they became more ideologically siloed and less cooperative, fewer and fewer bills were passed.Voting Network

On college campuses, homophily can lead to clusters of social networks struggling to engage with one another. Activists may mostly connect with other activists, athletes with other athletes, white students with other white students, and on and on. When complex campus discourse is required, the only way that all parties involved can learn from one another is if there are links between the many networks that allow information to pass from group to group. The more connected the various groups are to one another, the greater the chance that information held in one group will diffuse into another group. This is not to say that it is bad to befriend people with similar perspectives to our own. What this does suggest is that our social networks can use a little shuffling periodically. This shuffling process creates a more diverse network that comes with many interesting benefits.

Diversity is a buzzword on college campuses. Browse any college admissions website, and they’ll boast percentages of how diverse their student body is. What these pamphlets don’t mention is the wealth of research behind the benefits of diversity in networks, along with the unfortunate fact that many diverse perspectives often remain unheard because of a lack of connections between social networks.

Scott Page, an external professor at the Santa Fe Institute, has written in depth about the proven benefits of groups with diverse perspectives (diversity of academic perspectives, racial perspectives, experiential perspectives, etc.). For example, when solving complex problems, groups containing diverse perspectives outperform groups made up of the highest achievers (Page, 2010). In networks with a wide range of connections, “other regarding” behaviors such as empathy are more common (Grund, Waloszek, & Helbing, 2013; for a technical demonstration of this phenomenon showing a recent modeling experiment, see this video). People who engage with strangers on subways tend to report feeling happier (Sandstrom & Dunn, 2013, Schroeder & Epley, 2014). Introducing your distant friends to one another is associated with higher well-being (Anik & Norton, 2013). Even something as simple as rotating leadership in a group can increase innovation (Davis & Eisenhardt, 2011). On top of that, in groups that periodically rewire their connections, cooperation among the members increases (Rand, Arbesman, & Christakis, 2011).


Homophily happens. We like people who are like us. But what is also clear is that diversity and randomness in an otherwise rigidly-structured network can result in greater cooperation, innovation, and well-being for the people in the network. SwatDeck was our attempt to harness these benefits and to understand if, in a small way, we could change the social scene at Swarthmore.

On a sunny afternoon in April of 2014, the first SwatDeck groups disembarked on their adventures. I remember sitting at my phone for hours that day, obsessively swiping down, waiting desperately to see if there were any new pictures being posted. In a world where it’s so common to laugh off effort, to my surprise, #SwatDeck was being taken seriously. I saw pictures of unfamiliar nooks of Philadelphia, of students deep in conversation, of books, coffees, foods, dumplings, and of art. By the end of the month, we learned from survey data that each group spent between 3 and 6 hours in Philadelphia, sightseeing, wandering, eating, sitting. Many groups visited art exhibits. Some went to used bookstores. One group got a personal tour of a Buddhist monastery. The groups were lightly guided by discussion questions, although many groups didn’t discuss any of them and instead chose to share stories and insights from their own lives. At the end of the adventure, most students went their separate ways, with a new depth of understanding of their peers.

In all, 146 students filled out surveys before and after the adventure, with questions related to their demographic information, subjective well-being, academic interests, and extracurricular activities. We found that students reported feeling happier after SwatDeck, more connected to the average Swarthmore student, a greater sense of belonging on campus, and an increased willingness to donate to their alumni institution. Every student reported the experience as being worthwhile, and all but one wanted to do SwatDeck again.

SwatDeck attracted a majority of first and second year students (61%), as well as a slight majority of female participants (56%). The racial makeup of the participants was similar to that of the college as 45% identified as White, 24% identified as Asian or Pacific Islander, 14% identified as Hispanic or Latino, and 13% identified as Black or African American.

While certainly promising as a pilot, we are still refining SwatDeck’s experimental design to make it more rigorous. Some students didn’t show up, leaving a few groups of three instead of four. A few groups teamed up with other groups, forming an eight-person group. Quantifying each group’s relative diversity is difficult without extensive data on each student, and balancing the experimental component of SwatDeck with its organic and personal feel was no easy task.

What was clear, though, was that each student spent several hours becoming invested in the lives of a few other unfamiliar students. Some of the most encouraging data revolved around this fact, in the form of participant comments. Dozens of participants reached out to us with similar praise:

“SwatDeck was one of the most enjoyable and meaningful activities that I have done on this campus. I met people who I never had seen before, even when it feels like I know everyone at this small college, and I hope that I will continue to form those bonds… I felt like everyone had more license to be free and open about who they are… my experience with SwatDeck really reminded me how friendly, human, and caring people are.”


To see all of the pictures and comments from the students involved, visit swatdeck.tumblr.com or search #swatdeck on social media.

Over the next year, SwatDeck will be spreading to a few other college campuses, employing the same mindset of periodically injecting diversity into students’ lives. Decades of research, computationally, observationally, and experimentally, all say the same thing. Surrounding yourself with people who look and think and smell and feel the same way you do is selling short the vast potential of your intellectual and experiential growth. SwatDeck is an attempt to engineer a solution to this problem that is so easy to fall into. Of course, our whole lives need not revolve around this idea, but periodically, it is a breath of fresh air.

While much more could be done to improve the design and the experience of SwatDeck, it is nonetheless exciting to ask ourselves, what if groups with diverse perspectives are truly more innovative? What if we cooperated just a fraction more often? Are there any futures we could dismiss outright? Imagine a world with 0.1% more good ideas.

It might seem slight. But sometimes, all we need is that one idea. What if that business solution is sitting there in front of us, waiting for the right minds to connect to build it? Or that cure, or that invention? Projects that can increase the likelihood of making curious, distant connections are popping up in cities across the globe. Experiencing them really makes you think: What futures can we create together if we find new spaces to search for answers?

Brennan Klein

Picture 3Brennan Klein has written about cognition, language, and perception for The Psych Report. He will begin his PhD in Network Science at Northeastern University in the fall of 2015. Twitter: @jkbren


Further Reading and Resources


Tags: , , , , , , ,
0 replies

Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>