THE VIRGINIAN-PILOT
Copyright (c) 1995, Landmark Communications, Inc.
DATE: Sunday, May 7, 1995 TAG: 9505040487
SECTION: COMMENTARY PAGE: J2 EDITION: FINAL
TYPE: Book Review
SOURCE: BY ANN G. SJOERDSMA
LENGTH: Medium: 81 lines
A MATHEMATICIAN READS THE NEWSPAPER
JOHN ALLEN PAULOS
Basic Books. 212 pp. $18.
CELEBRITY mathematician and best-selling author (Innumeracy) John Allen Paulos' latest venture into American popular culture, promisingly titled A Mathematician Reads the Newspaper, would be more aptly titled A Mathematician Doodles While Reading the Newspaper. This frolic of a book is an idiosyncratic exercise in mathematical free-association - largely uninvolved with the practice of journalism. It frustrates, it confounds, it aggravates, it fascinates and, at times, it even entertains.
The probability that a hypothetically average student of math would read, finish and enjoy Paulos' arcane reverie is 1 in 100,000 - numbers that I, in emulation of the master, pull out of thin air for the sole purpose of having numbers to bandy about. Assuming that the current U.S. population is roughly 260 million, that means a mere 2,600 souls will relish this book - or does it? (Who, after all, is a hypothetically average math student?) The rest need only read the conclusion to glean the essential point: Be skeptical of all uses (read: nonuses, misuses, abuses) of math/statistics/numbers in the daily newspaper.
Those of us skilled enough in math to be skeptical, without Paulos' coaxing, of daily-newspaper number-crunching, however, want more than facile attitudinal advice; we want exposure and debunking. We want our esteemed, albeit publicity-seeking Temple University mathematics professor to horse-whip, figuratively speaking, all numbers offenders who manipulate public opinion with deceptive statistics and faulty mathematical reasoning. Intellectual revenge would be so sweet.
Alas, we do not get it.
Oh, Paulos, a newspaper junkie curiously enamored of headlines, satisfies us somewhat: He illustrates the (dis)advantages of using absolute numbers vs. a rate of incidence; expounds on means, modes and medians and their relevance; contrasts ``correlation'' with ``causation''; embellishes the concept of ``conditional probability''; exposes margins of error. But he spends most of his time supplanting reality with theory. Newspapers, oddly absent by name, source and example in this book, serve merely as a springboard into Paulos' mathematically skewed mind, a mind that travels easily from the newspaper fashion pages to a breakfast of squeezed and folded toast with jam via the concepts of unpredictability and deterministic chaos. (I think you have to read it to ``be there.'')
Consider his discussion of DNA testing, a topic much in the news with the O.J. Simpson case. Concerned with the ``conditional probability'' that a person is innocent given that his DNA imprint matches that of a crime-scene sample, Paulos eschews medical science to posit an absurd numerical argument. He assumes that in 2001 in a given city of 2 million people, all residents will have their DNA prints on file and further that in a given murder case three of them will have matching prints. What is the probability that one of these people is innocent? Why, two in three, of course, ``more than reasonable doubt,'' says Paulos. Would that reality were so tidy.
Or consider Paulos' mathematical breakdown of a political campaign in which he argues that single-issue (gun control) voters determine an election, despite their much smaller numbers: In assigning fixed percentages to the depth of feeling that voters have about a particular issue - described as a ``litmus-test'' issue - he ignores the many other factors that voters weigh. Some decisions - messy human ones - cannot be quantified.
A social critic operating within mathematical models, Paulos has a unique way of thinking, and of expressing those thoughts, that both compels and repels. A Mathematician Reads the Newspaper is chock-full of good, but weakly stated points about lapses in newspaper analysis (statistics are routinely unchallenged, for example) and manipulation of statistical samplings and conclusions, as well as about such myriad matters as quotas and voting blocs, truth in advertising, randomness and nonlinear systems; but its overall chaotic, off-the-wall quality detracts significantly.
Trust me: Only 1 in 100,000 will want to play this game. MEMO: Ann G. Sjoerdsma is book editor of The Virginian-Pilot and The
Ledger-Star. by CNB