Nearly 50 years ago, a computer engineer named Paul Baran peered into the future of American media and didn't like what he saw.
"With the diversity of information channels available, there is a growing ease of creating groups having access to distinctly differing models of reality, without overlap," wrote Baran, a co-founder of the California-based Institute for the Future and a pioneer of the early Internet. "Will members of such groups ever again be able to talk meaningfully to one another? Will they ever obtain at least some information through the same filters so that their images of reality will overlap to some degree?"
This was 1969. Baran was lamenting how the rise of television would cleave the political public. But his warnings may be more prescient today than ever: New findings based on an extensive survey of American book-buying habits find that readers on different sides of the political aisle are not only deeply polarized over scientific issues—they also read completely different scientific books.
"It's really a consumption divide," says James Evans, a sociologist at the University of Chicago and lead author of the study, which was published this week in the journal Nature Human Behaviour. "It's very difficult to imagine consumers of science in this environment appealing to a shared body of claims and facts and theories and arguments because they're really looking at different things."
Evans has long studied the history of science, and how scientists collaborate with industry. But recently, a conversation with Cornell University computational social scientist Michael Macy left him wondering whether the U.S's increasingly polarized politics would be reflected in how people view and read about science. The pair decided to team up to measure this polarization in a unique way: through the books they buy.
Unlike the more commonly used method of surveys, book-buying data is potentially more useful because it allows for much larger sample sizes, Evans says. Plus, it’s more anonymous than a survey: The books are purchased privately online and shipped in nondescript boxes to people's homes, meaning there's no fear of judgment from a pollster (a factor that may have helped skew polls before the 2016 U.S. presidential election).
Finally, purchasing a book requires a financial investment that makes it more likely that people are really committed to the view of that book, Evans says. As he puts it: "Talk is cheap. But if they're putting their money on the line ... this says they have a certain level of interest."
Evans and his collaborators drew on data from book giants Amazon.com and Barnes and Noble, which together have access to more than half of the world's book-buying market. They didn’t collaborate with either company, meaning they didn’t have access to buyers themselves. However, they were able to take advantage of a feature both websites offer: book suggestions.
When a customer buys a book from either site, a list of books that other people who bought that book tend to purchase will pop up. These suggestions "allowed us to build an entire network representation of that book-buying space," Evans says, linking hundreds of thousands of scientific books to each other in a web, along with more than 1,000 conservative and liberal books. All told, the team sorted through metadata for some 1.3 million books.
Researchers looked at that web to see what books about science are most often purchased by people who buy other books with liberal or conservative political slants (for example, a book by Rachel Maddow versus one by Ann Coulter). What they found was a stark divide in the kinds of science these two groups like to read about. Liberal readers more often picked books about basic science disciplines, such as anthropology, while conservative book purchasers tended toward applied science books, such as medicine.
"It's not just that they purchased different books, they purchased very different books from different regions of the scientific space," Evans says.
There may yet be hope for some measure of bipartisan unity. A few disciplines appeared to attract relatively equal interest from both sides of the political spectrum—namely, veterinary medicine, archaeology and paleontology. "Apparently we can all agree that dinosaurs are awesome," says Evans.
For science lovers dismayed by recent restrictions on the use of science at government agencies, there is another silver lining to the results: Political book purchasers of both persuasions were more likely to purchase books about science than topics like art or sports. "There's a really broad acceptance of the value of science,” Evans says, “by liberals and conservatives.”
The scientific fields that appeared most polarized among liberal and conservative-leaning book buyers may not surprise you: climatology, environmental science, social science and economics, among others. (By "polarized," the authors mean that there was very little overlap between what climate science books liberals bought versus the ones that conservatives bought.)
Evans worries that in the long-term, this polarization could not only influence how the public views science, but could shape science itself for the worse. "The concern is that this kind of polarization could end up shaping the production of science in those fields," Evans says—for example, leading scientists to design narrower studies that unconsciously seek to confirm results that align with their biases.
In an opinion piece published alongside the study, Georgia State University political scientist Toby Bolsen writes that the results underscore a growing concern about Americans associating themselves more with people and media with whom they share opinions on science and politics—which often leads to those opinions being strengthened. "This can impede science’s ability to enhance the quality of political debates," writes Bolsen, who wasn't involved in the research.
He cautions, however, that this study did not draw on a random sample of conservative and liberal books—they were picked by the researchers based on Amazon’s categorization of them. Nor does it address the motivations that drive an individual to buy or read a certain scientific book.
James Druckman, a political scientist at Northwestern University who studies how people form political preferences, says Evans' research is "clearly is a critical advance in what we know." Druckman, who also wasn't involved in this study, says the work "gives a much more nuanced and likely accurate view of partisanship and science." At the same time, he adds, "it avoids simplistic portraits of partisans."
This is far from the first effort to analyze so-called “information silos” using data. In 2014, when waves of violence were rocking Israel, data analyst Gilad Lotan published an analysis of the social media and news coverage of an attack at a school in the Gaza Strip. In a series of stunning maps, Lotan detailed the wide gap between the kinds of news outlets, posts and articles shared by those considered to be "pro-Israeli" and "pro-Palestinian" on Facebook and Twitter.
“A healthy democracy is contingent on having a healthy media ecosystem," Lotan wrote. "We need to be more thoughtful about adding and maintaining bridges across information silos online.”
In the future, Evans hopes to be able to work with online book publishers to collect specific data about buyers and their preferences. In the meantime, though, he hopes to see more work to bridge this scientific gap. For instance: scrutinizing book-recommendation algorithms to make sure that they don't box people into certain viewpoints, getting scientists to better communicate when there is consensus opinion in their fields, and creating more forums for people of different political views to discuss science.
"Doing that could allow us to make science a shared resource," Evans says. "I think the onus is on us as a society to grapple with this."