J.L. Wall fears that it will limit exposure to the classics:
Without the physical store, without the displays, the tables, the deals, the promise of no additional critical essays, what will guide a reader (no matter their age) to Herodotus, Thoreau, or Willa Cather outside a class syllabus? There’s great promise in the possibility of out-of-copyright works becoming more accessible to more readers than ever before—but there’s also danger in the thought that there will be nothing to guide a reader toward them by chance encounter, that they might come to smell even more of musty classroom learning than before.
Wall’s fears would not be assuaged at the recent Digital Book World conference:
“Look at a book as a bag of words,” suggested Matt MacInnis, another panelist, who had been working on education projects at Apple before forming an interactive-book company called Inkling. “Bag of words,” he pointed out, is a computer-science term: a model by which a machine represents natural language. “Computers are terrible at natural language,” he said. “Humans are shitty at multiplication and division.” For a reader searching the Internet for information, he explained, “the word rank is going to be terrible for a bag of words of book length.” But a book that is broken up into component parts would show up higher in an online search result, because each discrete section coheres around a single idea, which can be tagged, indexed, and referenced by other sites. This is known in the business as “link juice.”