Homophily: Center for Spatial Research

Wednesday, September 18th 2019, 12:00 pm
Chicago Cultural Center 3rd Floor

Paprika! in conversation with Laura Kurgan
Homophily: The Urban History of an Algorithm, 2019
Installation Description: LED light displays driven by an algorithm, explanatory graphics and data visualizations on digital wallpaper, documents from Paul Lazarsfeld and Robert Merton’s archives

Paprika: 00:11 We were thinking that we could talk about the project you are displaying here and afterwards hear a bit about your work with Center for Spatial Research. How does that sound?

Laura: 00:43 Yeah, sure, sure. Okay, great.

Paprika: 00:52 You decide where we start.

Laura: 01:19 The best place to start is on the outside. In the context of the biennial we have done this exhibit consisting of five two-sided walls. On the outside it’s displaying the algorithm that we discuss on the inside. As you can see the display starts with two-colored lights randomly organized on all five panels. By an algorithmically driven program based on the concept of the show – which is called Homophily – you notice that the lights are moving to where they feel comfortable, which is surrounded by lights with the same color. It turns out that by the time the lights stop moving they are segregated into groups of the same color. So it’s an extraction. It’s a pattern. What we do on the inside of the exhibit is to unpack the algorithm. It’s kind of like lifting the hood off of the black box. Not completely because we don’t fully understand the complicated math that drives these things in places like social networks. But we’re doing it in a very conceptual, simpler, abstracted way because it’s the concept which is really important in this context.

Lets go inside. We expose the origin of the term Homophily which was first found in a paper published by two sociologists in 1954, Paul F. Lazarsfeld and Robert K. Merton. They were studying a biracial housing project in Pittsburgh, Pennsylvania in which the residents would decide to divide into white residents and black residents living in different buildings. Lazarsfeld and Merton did a survey and to define the term Homophily they only used two out of 19 questions that they asked the residents: “Do you think white and colored people should live together in housing projects?” and “On the whole, do you think that colored and white people in the village get along pretty well? Yes or no?” So there were 487 people living in this housing project. It turned out that the black population were all liberal and they all thought it was a good idea, which is not surprising, but the white population had a much more varied response. Based on the white population, they counted what’s called “over-selection” to define the term Homophily. So it’s kind of an irony of history that these were questions that were asked by the sociologist and then it became a naturalized concept on social media much later. This section is a very important part of it, as it shows how the black population was ignored.

We look into the first questions that the sociologists asked. They would try to ask: “Who are your three closest friends?” That very much prefigures Facebook and other social media, so we kind of moved right forward into Facebook where we are asked to choose our 500 closest friends. The question is, when we do that are we ’doing the homophily’? Do birds of a feather flock together and does similarity breed connection? Which means: do you only become friends with people who you share the same values with? Say if you had in your network a 67% tolerance for being friends with people who are unlike you, then the Facebook-algorithm-switch will send you recommendations and advertising that would most probably stay integrated, right? If your friends are mostly like you, which means that you have a much lower tolerance for friends who are unlike you, the Facebook algorithm exacerbates segregation, which is for the most part what happens, which is why you get all this polarization and filter bubbles.
We look at the two original publications of the term. There is an unpublished version and a published version and the unpublished version asks much more institutional questions and doesn’t come down on such a precise methodology. But the way that the term gets picked up by computer science is by quoting this paper. So we looked at all the citations to see when this term gets picked up, and you can see how it happens massively when social media becomes big in 2005 to 2010 and most often in computer science and sociology which we visualized over here with a diagram called a “concordance”. It takes all the thousands of citations and shows you the context of the word. We have an online version of this, where you can click on a link to the actual paper. Lastly, we go into what became the inspiration for the outside walls. This is Thomas Schelling. He is an economist who studied dynamic models of segregation and in particular white flight out of the heart of the cities. He played a game with his students with pennies where they had to move them around and this is really how it became a model, a model of Homophily, and where it was applied first. Afterwards, [the model was] also applied in urban space but then moved to social networks where it’s most intensely deployed now.

Paprika: 11:18 Your discussion of Homophily makes me think of the notion of The Smooth City coined by Rene Boer in a recent article in Volume Magazine. He describes a tendency in urban development in which cities are scripted with “targeted” functions and users. That process creates smooth, frictionless, and segregated urban environments of homogeneity. Do you think that your study of Homophily – departing from an urban condition which is traced in social media – also points back to
the creation of specific spatial conditions today?

Laura: 12:08 It’s funny because I’m in an architecture biennial,
so everybody is asking that question, and I don’t think it’s the right
question. I think that what is on social networks has a huge effect on physi-
cal space, but it’s not a direct relationship. Smart Cities, they are deploying algorithms. So smart city logic tries to make these calculable cities
and it’s the same logic but it’s not about physical space; it’s still about this
overlap between the digital and the physical. It is in a way dumbing it down when you start trying to talk about it in terms of physical space
and how we develop cities and such. We have tried to dumb down the algorithm because there’s no way we can replicate it exactly. It’s not
a metaphor, actually. It is an extraction from physical space into network space. So that’s what we’re trying to show. A type of influence of archi-
tecture on network space. And then we’re going to move from here to “Weak Ties” and “Nearest Neighbor”. We’ve been collaborating with
Wendy Chun who is a theorist. She’s doing a big project on digital democracy and our bigger project is on spatial inequality. Here we are bringing these two things together and will do a lot more work together
on this topic.

Paprika: 15:08 Throughout time, more and more of what we do in physical space we are now doing in digital space, such as communication and leisure, everything is in digital space. So the fact that you see this transformation from physical to digital very much speaks about our
society today, and how we are more or less living in the digital. Do you have a preference of interest in digital over physical space?

Laura: 15:51 Oh no, not at all. No. I think they are just so linked. Yeah. I don’t think one space takes a priority over the other anymore. So that’s why when architects ask “what does it mean for physical space?” That’s the wrong question. It’s about how they intersect each other.
The intersection makes you think anew about both. You have to think about both to solve the problem.

Paprika: 16:24 Can you speak a bit about the work of the Center for Spatial Research that you are directing at Columbia University?

Laura: 16:35 We are doing mapping and data visualization but we don’t take any of those terms for granted. We always question the
power and the governing structures that are behind every map, whether it’s the census, traffic control, or the politics of what brought them
into being. A map nowadays might as well be called a data visualization
because we are doing digital maps which are drawn with datasets.
Our job as spatial experts, being in an architecture school, is to demap
and unclutter some of the presumptions that might rule those things.
This is Jia, she is a postdoc in the Center for Spatial Research right now and she has been part of the team. Particularly with these concordance diagrams and visualizations. If you have any questions on the visualisations specifically you can ask her. These are two students from
Yale School of Architecture who are doing an interview for the school
magazine.

Paprika: 18:00 Hi Jia, nice to meet you. May I ask a technical
question on how this concordance diagram
was generated?

Jia: 18:04 It is generated by doing a word search and then aligning the result with the word that we searched for which is in the center. It is an existing idea used for reading texts and in fact the origin of the word concordance comes from Biblical studies.

Paprika: 19:26 Thank you so much.

Laura: 19:29 Thank you. It was lovely meeting you.
Say hi to Keller from me next time you see her.

Paprika: 19:30 We will! Bye, see you at the opening party!