John Sack has probably done as much as anyone to transform—or disrupt, depending on who is talking—the scholarly publishing business. As founding director in 1995 of HighWire Press, Sack was asked by Stanford University Press to see if research journals could dip a toe in the unknown but rising waters of the online world while still keeping their integrity and citable permanence dry. Sack plunged in. Today HighWire is the leading e-publishing platform, handling 1,732 scholarly journals, reference works, books, and conference proceedings, for scholarly publishers, large and very small.
It has been an exciting ride, says Sack, but in recent years, HighWire has noticed something puzzling. It's not just the journals that have changed, he says. The readers have changed. Reading itself has changed. "We did a study in 2002 in which people told us that they read three or four journals regularly," Sack reports. "In 2012, we asked the same question. They told us that they read 8 to 10 journals."
The result threw Sack and his colleagues for a loop. "How can this be possible? People don't read faster. Evelyn Wood [a 1960s promoter of speed reading courses] was ultimately not successful. The journals aren't any smaller. The articles aren't any shorter. What was going on?" What has changed, HighWire discovered, is how people define reading. "What we found when we probed is that what people mean when they say they 'read' a journal now is that they read the email table of contents (ToC)."
More people skimming more ToCs has many implications for editors and publishers of scholarly journals but it is also a clue into the changing nature of reading. When Sack surveyors pressed scientific readers on what would make finding the content they wanted easier on ToCs and elsewhere in journals, the answer once again astounded Sack. "What people said was that finding content is easy. They were already finding too much stuff. Reading takes a lot of time."
Thinking of ways to make reading of ToCs and articles faster, Sack and his colleagues drew up a list of 20 possible innovations. Readers liked some of them—short titles, visual abstracts, "take-home" messages, and structured abstracts. Authors liked others, particularly impact metrics ("Authors always want to know how many times their paper is being read"), but nothing caught fire. Journal editors, on the other hand, were adamant about encroachments on their editorial mission by giving in to arbitrary lengths in titles or posting minute-by-minute metrics on individual articles. "They said, 'This makes it sound like this paper is more valuable than that paper.' They say, 'The hell with the horse race. I'm not going to encourage that'."
That sent the HighWire surveyors back to readers to find out what makes an article jump out of a ToC as a "Need to Read' paper. "What we found in our talks was that people were asking for different things and conflicting things. This was not making sense. Some people were saying, 'I need more original research data' while other people were telling us that they need more compact stuff: 'I need it summarized.' We thought, how is this possible?"
An insight came from a neuroscience graduate student who said he read on three levels, according to Sack. The insight wasn't from extensive field surveys or a randomized control trial, but says Sack, "it has helped us sort the evidence we were getting."
There are three levels of reading, says Sack, taking the part of the grad student faced with too much content. "On level one, there is one set of things that I will read because they are directly related to something I am working on where I am an expert in this field. As a result, I don't care what journal it's published in. In some ways I don't care who the author is although I may eventually care. But I'm an expert and I don't care if it's on a preprint server or what the editors of Science or Nature think. I will read it because I am an expert in that field.I can judge quality for myself."
Level Three is the other extreme, says Sack. The logic goes like this: "I will read it because I'm a researcher and I'm a scientist and I should be well informed about hot trends and what's going on outside my field. And here I look to Science, Nature, and Cell, and, according to some evidence, to the New York Times 'Science Tuesday' section." The reading problem is in the middle, says Sack. If you allow the two information poles 15% each, that accounts for 30% of content. The 70% in the middle is where readers need help. "And that's where people said that we need better tools and a better experience to read that stuff faster," says Sack. For HighWire, the current response is to find tools that help you, the scientific reader, plow through that 70% faster.
The future, Sack believes, will be the other way around: The content will find you. You'll still have some journals that you watch yourself but the concept of the personal journal means that even more advanced search algorithms will learn your reading habits and tailor content flow to your needs, interests, and whims. "That's the concept of the personalized journal, but whether you will create the personalization or clever people at Elsevier will create it, I don't know," says Sack. That question hangs on the evolving shape of the journal publishing business.
Preserving diversity in scholarly publishing was one of HighWire's founding goals, and Sack believes that it remains critical to the future of scientific reading. "I want to make sure that small scrappy publishers survive because they will innovate. Large publishers innovate, there's no doubt about that. Elsevier does some pretty cool stuff but the really unusual stuff will be coming from the little guys."
But Sack catches himself. If there's anything he's learned since HighWire put its first journal online in 1997, it's that making predictions is risky. For one thing, predictions now are searchable and retrievable. "That's the problem with predicting the future these days. Someone can easily go back and find out what you said 20 years ago."