April 23, 2015
Diversity leads to stronger outcomes, says lecturer
In 1650, around 30,000 published books existed. An avid reader could learn nearly everything there was to know at the time. Now, there are 1.5 billion published books and counting.
“In a small world it makes sense to think of someone knowing a lot more than another person. In the great, big world, people just know different stuff,” said Scott E. Page in his talk, “Ability and Diversity: The Academy and the Cognitive Economy,” on campus April 22.
Page, director of the Center for the Study of Complex Systems and a professor of political science and economics at the University of Michigan, said that in our large, complex society, diversity is often more important than ability.
“I want to argue that diversity – defined as differences in how we see the world, how we think about the world, how we try and solve problems, the analogies we use, the metaphors, the tools we acquire, the life experiences we have – make us better at what we do,” he said.
In a computer simulation featuring 15 different models, Page pitted two groups against one another to solve a difficult problem. The first “alpha” group consisted entirely of high-ability individuals, while the second was a diverse group of people with varying levels of competence. Under every model, the diverse group outperformed the high-ability group.
A mathematical formula exists for this concept, according to Page, where “crowd error equals average error minus diversity.” Therefore, to have a small crowd error, two possibilities exist. Either the crowd must consist entirely of people who know how to solve the problem – meaning the problem is easy – or in the case of a difficult problem, “have people who sort of see through different lenses, all of which are good, that are collectively different.”
Countless real-life examples of this phenomenon exist, too, said Page. In 2006, Netflix launched a contest offering any team $1 million if they could invent a 10 percent better version of their movie-predicting model, Cinematch. After years of working on the project to no avail, the team in first place decided to combine with the second- and fifth-place teams. The resulting team’s model bested the 10 percent goal, inspiring other teams to combine, too. A second team also ultimately made it over the 10 percent mark.
Page said studies looking at authors of research papers and patents show similar results. For instance, the chance of writing a successful paper, defined as one that receives more than 100 citations, increases more than fourfold if it’s co-authored. However, Page said, it’s not the number of authors that counts, but the different ideas they bring to the table. Co-authors of similar ethnic groups received fewer citations, while co-authors from different universities had more, according to Page.
Yet he said that some of the worst ideas come from diverse groups, too. Diverse groups tend to have less confidence in their abilities than homogenous groups, and they often disagree more. Therefore, Page said, it’s imperative that people develop the ability to work well in diverse groups: “Because of the fact that talent doesn’t know color, or gender or race … it’s super important that people develop this skill set. You just can’t put the ingredients in a blender and hope it’s gonna work out.”
“One reason why you want to bring in a diverse student body and a diverse faculty is you want to train the next generation of people who have to solve these hard problems how to be in groups, how to engage, how to interact in a multiracial, multiethnic, multigender setting,” he concluded.
The annual Robert L. Harris Jr. ADVANCEments in Science Lecture is sponsored by the Office of Faculty Development and Diversity.
Natalie O'Toole '16 is a student writer intern for the Cornell Chronicle.