The dancers provide the code, and the computer scientists transform it into choreography. That’s the basic idea behind a collaborative project between Assistant Professor of Dance Charlotte Griffin’s Advanced Dance Composition class and Associate Professor of Computer Science Stephen Majercik’s course, Nature-Inspired Composition.

Dancers in Griffin’s class combined different movements, each assigned a number, to create performances that can be reduced to numeric sequences. These sequences constitute a code for the computer science (CS) students, who write “fitness functions” that determine which sequences are best (although the dancers bristled at the idea of one dance being objectively better than another). The function—through a process that Majercik compared to evolution—will spit out a new sequence of movements, a type of computer-generated choreography.

As often happens when two disciplines come together—especially disciplines as disparate as CS and dance—there have been some slight hiccups. Griffin said that the project has involved a departure from the typical language of dance.

“A lot of this language is very challenging from a choreographic standpoint, like picking the best, ranking in order of preferences,” said Griffin. “In choreography I’m not using words like good or bad. It’s more like interesting or contrasting or complementary.”

Majercik has had to adjust his vocabulary as well. He admitted that his dance experience is limited to social events like weddings and parties.

“I don’t think I’ll make this mistake again, but I made the mistake several times of saying we want something good,” Majercik said. “And [Griffin] says, ‘No, we don’t want something good. It doesn’t have to be good—it can be interesting.’”

Only a few minutes later, Majercik made the same mistake again while describing the challenge of building a fitness function to evaluate the dances.

“Here’s the crux of it. Here’s where the difficult piece is,” he said. “For the evolution to work there has to be a function that evaluates the sequence and gives it some sort of score—how good is it?”

“Or how interesting,” Griffin interrupted, prompting laughter from both professors.

Dancers in Griffin’s class also struggled to adapt to the language and logic of CS, particularly when students from Majercik’s class visited the studio to rank the dance sequences and develop criteria for their fitness functions.

“I don’t believe in ranking dances like that, especially among people that you’re in class with,” said Sarah Guilbault ’18. “It’s not a quantitative thing to rank dances, but it’s necessary for computer science in order to create new sequences.”

Lucy Saidenberg ’15 had similar reservations, but found that the project was still valuable.
“It was a little scary at first, but I think the concept was really interesting in the end,” she said.

Both dancers and CS students underscored the potential awkwardness of interdisciplinary collaboration when they described the meeting between their classes. Simon Moushabeck ’16, who is in Majercik’s class, said that “it was like studying another species” and Saidenburg joked that it would be “hilarious” to see the CS students dance.

Griffin and Majercik may seem like an odd pair, given the enormous differences between CS and dance, but they share the belief that their respective fields can benefit from a carefully designed collaborative project.

“We’re working on theme and variation this semester, so this is an opportunity to look at movement analysis and movement variation in a way that a choreographer might not do on their own,” Griffin said.

Majercik said he was excited by the questions and uncertainties of the project.

“It’s been really interesting trying to figure out how these two processes can work together,” he said. “Can you think of it being a collaboration between a dancer and the algorithm? It’s really an exploration. We’re not sure where it’s going to lead.”

Griffin and Majercik began planning a collaborative effort two years ago, when they served on a working group as part of the College’s Digital and Computational Studies Initiative (DCSI). Their current project is not a formal part of the initiative, but they think of it as an experiment in what they call the “computational arts.”

The two professors see this semester’s project as a trial for a future course in which students would participate as both dancers and computer scientists. They would call that course Computational Expressivity.

Majercik said that a number of students in his class seem interested in such a course.

Moushabeck said that he has enjoyed applying CS principles to a real-world problem.

“I think it’s a cool idea,” he said. “It gives us the chance to put something that we’re learning to use, to see it make results in a way that’s not just seeing the numbers fly down on the screen.”

Not all the dancers were as sure about the idea, however.

“I don’t love the idea of a computer telling me which moves are the most aesthetically pleasing,” said Saidenburg. “I don’t really believe that it’s possible, just because of the subjective nature of what’s good dance.”