An Analysis of Coherence

The brief essay on this page, "Accounting for Taste," was written by James Gleick, former editor for the New York Times, lecturer at Princeton, and author of three books about how technology affects our lives. (Two of those books were Pulitzer Prize and National Book Award finalists. Excerpts from his most recent book, Faster: The Acceleration of Just About Everything are available online.) "Accounting for Taste" is used with Mr. Gleick's permission.

Using blank text blocks, we have divided Gleick's essay into several parts. We want you to analyze how the language above each blank is held together by the four mechanical devices for coherence: pronoun reference, repetition of key terms, transitional tag words, and parallel form. (Click HERE to review those concepts.)

If you would like to read Gleick's essay without being interrupted by the blank text blocks, click HERE. (There are hyperlinks to several of James Gleick's essays on that page.)

Hello, James Gleick," said Amazon.com the other day (click here if you’re someone else). "Take a peek at your brand new music recommendations."         I peeked. Amazon’s computers predicted that I would like the Beastie Boys, Adiemus, Frank Sinatra, Harvey Danger, and the Dave Matthews Band. What an impressive list! All right, I don’t actually care for any of these, but still. It was quite a shot in the dark, considering I’d never been to Amazon’s music department before. This is the way it’s going on the Internet: if marketers want your money and your time and your "eyeballs," they feel they should figure out who you are and what you like.         Not only does their software try to calculate your taste in music by keeping track of the music you buy, it even tries to work out your taste in music from your reading habits. This could be a parlor game: If you like Vladimir Nabokov, maybe you’ll also like Igor Stravinsky? If you like War and Peace, maybe you’ll like the 1812 Overture?         If you like E. L. Doctorow’s Ragtime, maybe you’ll like Scott Joplin’s ragtime?         One Flew Over the Cuckoo’s Nest and 10,000 Maniacs?         Consumer Reports and Crash Test Dummies?         Kafka’s "Metamorphosis" and the Beatles? We like to believe that our souls are our own and there’s no accounting for taste. So it’s disconcerting to find that, on line, there’s suddenly lots of it. Amazon has its BookMatcher, the music store CDnow has its Album Advisor — sooner or later every merchant of just about everything will follow suit, analyzing your private likes and dislikes with "real-time recommendation engines" based on "collaborative filters" and fuzzy logic.         The basic idea is the same everywhere. Say you favor turtlenecks, convertibles, nautical history, bebop, and zinfandel. No doubt you are proud of your rare good taste, but it’s a big world, and somewhere in a million-entry electronic-commerce database are a few other people with the same preferences.         If you knew that your doppelgängers were raving about the latest Mike Leigh movie, wouldn’t you want to give it a look? In the jargon of the collaborative-filter game, these weird pals are your "community" and your "trusted associates." Their taste might be more in tune with yours than the few people you trust in your own small circle of friends.         The whole thing is just a mildly clever database look-up, but maybe it works, at least for some people and some kinds of taste. It has no intelligence about the content of the merchandise — Mozart and Madonna might be flavors of ice cream, for all it knows. It only has the beginnings of what could become a formidable electronic dossier: your purchasing history plus your volunteered comments about what you love and what you hate. At CDnow, for example, a customer can choose buttons for "I own this already" or "Not for me"; the computers, of course, watch and learn.         It’s scary. "Is one’s entire psyche’s most secret landscape really a fairly public thing, given just a few pieces of information?" asks Douglas R. Hofstadter, the cognitive theorist and author of Gödel, Escher, Bach. "If you know that I love Chopin and Bach and am totally cool to Beethoven, can you predict that I love Cole Porter and Fats Waller but am indifferent to Oscar Peterson and Charlie Parker, and hate Elvis Presley?         "What is disturbing, to spell it out, is the idea that one’s taste, which seems like such a personal thing, connected with and determined by one's inmost being, should have, in a way, a mechanically, nearly deterministically, knowable nature." Maybe we’re not quite knowable at that. And these are computers, so the mistakes they make can look very, very stupid. When they go off the rails in a sensitive area like taste, some people get angry. "The worst thing of all," says one irate customer, Russ Korins, "if you like Third Eye Blind — who sing ‘Semi-Charmed Life,’ that song that goes doo doo doooo, doo doo DOO do, and ‘Graduate,’ a mainstream version of Avenue A rebel rock, screaming, ‘Can I graduate?!’ — then what do they also recommend? The Four Seasons by A. Vivaldi."         At CDnow, they take this sort of thing in stride. "You can’t argue with the customer — they know what they like," says Evan Schwartz, director of product management. "People love to click on ‘Not for me,’ ‘Not for me.’"         When customers click, and especially when they buy, they add to the storehouse of information about that mercurial, irrational, chaotic thing we call taste. As knowledge builds up, maybe the computers will stop recommending Vivaldi to Third Eye Blind fans. Or maybe it will turn out that Vivaldi and Third Eye Blind have some kind of century-bridging affinity, even if no musicologist could say exactly what. Or maybe it’s a moving target, and last year’s Third Eye Blind fans have a different sensibility from this year’s.         You might think of these growing databases as merchandising dossiers. Marketers are keeping a file on you, and if it’s not as tangible or incriminating as your F.B.I. file, it’s too personal for comfort. "These exhaustive lists become much more than mere lists; they act as electronic psychoanalysts," writes David Shenk in his recent book, Data Smog.         Under pressure from privacy advocates, most companies in the collaborative-filtering business pledge not to share information without customers’ consent. Even if the trail of your reading history leads your bookseller to conclude that you’re on the verge of buying a new red Porsche Boxster or a blue Gap dress, Amazon promises not to tell the car companies or the special prosecutor.         Still, if we have learned anything, we know that information tends to get around. Do I really want the whole Web to know I’m a Beastie Boys, Frank Sinatra, Harvey Danger kind of guy? I didn’t even know that myself. Name: Email address: Class: NOTE TO THE STUDENT: If the answers are sent successfully, you will see another page come up in the web browser. If you don't see this page, it is possible that an error occurred during transfer and you should either resubmit your answers or notify your instructor. Generated by QuizMaker 2.0. QuizMaker 2.0 for QuizServer © 1998 University of Hawaii. Developed for the University of Hawaii Office of Technology Transfer and Economic Development in cooperation with Maui Community College. All rights reserved. Any copying, distribution, or preparation of derivative works is strictly prohibited.