The future of text
The New Yorker still does great prose:
The Google Library Project has so far received mixed reviews. Google shows the reader a scanned version of the page; it is generally accurate and readable. But Google also uses optical character recognition to produce a second version, for its search engine to use, and this double process has some quirks. In a scriptorium lit by the sun, a scribe could mistakenly transcribe a “u” as an “n,” or vice versa. Curiously, the computer makes the same mistake. If you enter qualitas—an important term in medieval philosophy—into Google Book Search, you’ll find almost two thousand appearances. But if you enter “qnalitas” you’ll be rewarded with more than five hundred references that you wouldn’t necessarily have found. Sometimes the scanner operators miss pages, or scan them out of order. Sometimes the copy is not in good condition. The cataloguing data that identify an item are often incomplete or confusing. And the key terms that Google provides in order to characterize individual books are sometimes unintentionally comic. It’s not all that helpful, when you’re thinking about how to use an 1878 Baedeker guide to Paris, to be told that one of its keywords is “fauteuils.”
Taken from a story about the future of text by Anthony Grafton.
Lately I am becoming more and more frustrated by search, and turn to browse where possible. I’m not always satisfied to allow a computer algorithm to present what exists, because as Grafton points out, search misses things (and also turns up wholly irrelevant results). Drilling into hand-picked categories often provides a survey of available information that reveals unexpected links and sources. This approach may seem antiquated, but it’s still enlightening.