In January 2015, multimedia artist and web developer Brian Foo posted the first track in his year-long Data-Driven DJ project, an endeavor billed as a series of “experiments” rather than compositions that combine “data, algorithms, and borrowed sounds.” His working method allows him to mix areas he is at home in—namely computer science and data analysis—with a skill he hopes to learn: how to make music.
For his initial outing, Foo used census data to chart the rise and fall of median income levels along a subway line in New York City. He also selected samples from the work of resident musicians. He then wrote a computer program that combined those two elements to produce his final (after many iterations—more on that in a bit) sonic representation of the information. It sounds like this:
If you know the work of Steve Reich, it is impossible not to draw an immediate sonic connection here—and that is in some sense very much part of the point. Foo used a carefully selected palette of raw materials to conjure a fuller mental image of what, on the page, is rather a drier set of facts and figures. And he approached this challenge by using elements that he felt authentically connected to the given subject matter. Reich’s New York Counterpoint and the technique of phase shifting also spoke to Foo’s interest in exploring his own understanding of musical concepts while developing the listener’s emotional investment in the matter at hand.
“You’re never really told how to feel when you look at a chart, even though it may have to do with something that is very human like income inequality or climate change,” Foo explains. “It ends up being just information that you don’t really internalize or contemplate. So I wanted to figure out how to take something like a chart but curate an experience in which you’d feel a certain way while listening to the song.”
He also wants to take advantage of another aspect of music: its ability to grab your ears and not let go. “You don’t really think about a chart all day after you see it, but with music, it kind of repeats. So if I could embed information into the music, then those topics and issues would circulate in somebody’s head.”
So far he has applied his data-driven approach to music making to an engagingly diverse selection of topics: seizures, smog, romantic attraction. For one track, he developed a complex methodology to analyze the relationship between painters Lee Krasner and Jackson Pollock through samples of their work. The next month, he was tracing the movement of refugees over the course of four decades. Despite his desire to curate an experience for his listeners, his subject matter has also forced him to confront some interesting questions about the relationships between sound and emotional manipulation: Should the music sound “better” when passing through rich neighborhoods in his presentation of income? Should the sounds be uncomfortable during the active period of seizure? Where is the line between representation and judgment?
Foo says he’s had the most aesthetic luck with data that follows a curve, thereby supporting a climax and resolution within the track. His process also involves a certain element of ongoing surprise as he combines sounds and iterates the track over and over (and over) again until its parameters have been tweaked to the point that it meets his sonic desires. But he won’t artificially manipulate the results to improve only a certain section, requiring instead that the full track must remain true to the data from start to finish. Yet tweaking at the level of the algorithm changes the entire song, so it can be a frustrating process of trial and error to land on the final result. “I’m like a mad scientist just trying to mix different things together until something happens,” Foo admits. “So it’s almost a brute force process where I just iterate hundreds, if not thousands of times, until it sounds good.”
If you’re curious at this point precisely how these data sets end up being represented by sound, well, the details are only a click away. Core to Foo’s project is complete transparency when it comes to process. His tools and methods are meticulously logged on the project website for each individual piece, and if visitors find a way to expand, modify, or improve on his process he hopes to hear from them. This fits comfortably with Foo’s larger philosophical tenets concerning making art more accessible and inter-personal. He even offers fans the option to pre-order the eventual album by trading sounds with him or promoting the project in lieu of cash payment.
Foo suspects that he invests one to four hours every day into making each month’s track, and, contrary to expectation, his working process is actually lengthening—likely attributable to the increasing complexity of each sonic experiment as he acquires new skills and experience.
And that is his preferred way of learning, he concedes. “I try to not learn correctly. I basically try to avoid learning the right way for the longest period of time as I can because I think there’s this unique point in time when you’re learning that you don’t really know what the limitations are. When you don’t know the right way of doing things yet, you don’t really know where the ceiling is, and I wanted to see how far I could get just with the things that I already knew.”