The computer-related stuff is in the middle this month, starting at the fourth subhead. The math is all up here at the beginning, and the Latin is at the end. The controversial stuff, like affirmative action and presidential politics, is on the first two pages, the Christmas-shopping advice appears just past the halfway point, and the importance of anagrams is revealed in the last paragraph.
Hope that helps you find what you want.
When a book has a title like A Mathematician Reads the Newspaper (Basic Books, 1995), you make certain assumptions.
John Allen Paulos has written two earlier books on the misuses of mathematics--Innumeracy (Vintage, 1990) and Beyond Numeracy (Vintage, 1990)--so I expected his latest to be all about the mathematical blunders and blinders of the popular press. Since there's so much material to draw on, I expected a book full of outrageous howlers and wondered why it was only 212 pages long.
As it turns out, the book might be better titled A Mathematician Muses about Newspapers and Other Topics Suggested to Him when so Musing.
Paulos is a newspaper junkie. He grew up loving newspapers, and today he subscribes to The New York Times and the Philadelphia Inquirer; regularly skims the Philadelphia Daily News and The Wall Street Journal; and often reads The Washington Post, the suburban Ambler Gazette, the Bar Harbor Times, "the local paper of any city I happen to be visiting," various scurrilous tabloids, and even USA Today. Oh, and he is also a mathematician.
He has, he says, organized the book like a newspaper. I don't challenge that, but I do note that when John Paulos organizes a book like a newspaper, it looks a lot like when Marvin Minsky organizes a book like a mind (The Society of Mind, Simon & Schuster, 1988).
The chapter titles suggest topical relevance: "Clinton, Dole in Sparring Roles," "Cult Members Accuse Government of Plot," "DNA Finger Murderer," "Cellular Phones Tied to Brain Cancer," and "A Cyberpunk Woody Allen." Sometimes the chapters live up to the suggestion.
For example, in the chapter "761 Calories, 428 Mgs. Sodium, 22.6 Grams of Fat per Serving," Paulos points out the meaningless precision in news stories and recipes. A recipe, unlike other kinds of algorithms, can often get away with specifying "a pinch of this, a dash of that." But it should not then claim--as Paulos says they often do--to yield, say, 761 calories, 428 mgs. sodium, and 22.6 grams of fat per serving. In fact, the variations in the nutritive content of growing things and in the size of "one medium potato," for example, is so great that probably nothing you say about a recipe has more than one digit of precision. Meaningless precision occurs elsewhere, of course: Paulos tells of his neighbor, who brags of getting 32.15 miles per gallon, and of his daughter's teacher, who gives her a grade of 93.5 on an essay.
In the chapters "Lani 'Quota Queen' Guinier" and "Tsongkerclintkinbro Wins," Paulos discusses the mathematics of voting. It's fascinating stuff.
In the latter chapter, he presents the inconclusive results of the voting for five candidates in a mythical state's Democratic caucus, then presents five plans for a runoff election. All five arguments are plausible, and each leads to a different winner, demonstrating that "fairness" is a slippery concept.
In "Lani 'Quota Queen' Guinier," Paulos introduces other subtleties of voting. Consider a corporation in which the three stockholders have 47, 44, and 9 percent of the stock, respectively. If you think about it, you'll see that all three stockholders have equal power. Now consider four stockholders with 27, 26, 25, and 22 percent of the stock. If you consider all possible coalitions that produce over 50 percent, it becomes clear that the fourth stockholder's 22 percent wields zero power. The same kinds of disenfranchisement can occur in politics if voters vote as a block, due to racial antagonism, for example. Various innovative voting schemes have been proposed to get around this unfairness. Ms. Guinier was one of the proposers, and she fell afoul of a press and public that didn't understand the mathematical justification for considering such proposals.
One last voting example: Although the nine Justices of the Supreme Court decide all issues democratically, by majority vote, Paulos presents a scenario in which three Justices could control all decisions of the Court. Assume that five Justices, say the five most conservative ones, agree to vote together to ensure that their majority always rules. To accomplish this, they meet secretly to decide how they will all vote. They may not be in complete agreement, so they reach this consensus by the obvious method: They vote, agreeing to be bound by this prevote vote when the real vote comes in the full Court. Now assume that three of these five Justices hold an earlier meeting in which they decide how they will all vote in the prevote vote. These three, being a majority of the five, will carry the prevote vote, and their will will prevail in the final decision. And if two of the three, say Scalia and Thomas, hold an earlier meeting....
Relevance? As I was reading the book, the television was murmuring in the background, keeping my subconscious supplied with a steady stream of O.J. Simpson references. At one point, a DNA expert testified that a particular DNA pattern will occur in only one out of 57 billion African Americans; let's see, that's 3000 times as many African Americans as presently exist.
I also saw Senator Phil Gramm on CSPAN-2 defending the supermajority voting rule for getting cloture in the Senate. "That [sic] what we call democracy," said the Senator, showing that, while he may have the concept, he's a little loose on the terminology.
Paulos has a job for life rooting out innumeracy. As I was writing this column, population geneticist and prosecution witness in the O.J. Simpson trial Bruce Weir testified that when you look at the number, you see that the concept of race really has no meaning. Meanwhile, my inbox holds a questionnaire from the county of Santa Cruz pertaining to jury duty, and one of the questions asks for my race.
Or this one: Dr. Weir made an unwarranted assumption in one of his calculations, so he recalculated his result without the assumption. The defense correctly pointed out that the new expected value, having gone down, was more favorable to the defendant. The prosecution then tried to use the fact that the top end of the confidence interval had moved up to claim that the new results could be seen as less favorable to the defendant. Weir, being a good and honest mathematician, didn't let the prosecutor get away with that one: Both ends of the confidence interval had moved out--the top end up and the bottom end down--which meant that the expected value was less reliable than he had formerly reported.
If Monica Berg hadn't called to ask where this column was, I could have come up with a dozen more examples of innumeracy.
Some of Paulos's topics are, well, interesting, but not what you'd expect; for example, the little essays on self-reference and information theory that have only vague connections with newspapers. And the list of the top-ten reasons we love top-ten lists. (Number 10: "People...like to see if it's going to run out of good points before it gets to 10.")
The quirky asides are entertaining, and there's plenty of the germane stuff. He presents a rational look at touchy topics like racial balance in hiring, the meaning of SAT scores, and risk assessment with respect to firearms, abortion, and smoking.
The President of the United States, a former state governor, is big on finding out what works in state government and proposing that the nation do likewise. Hawaii's health plan is a great success, so why not use it as a model for the nation? Paulos explains why not: Sometimes things don't scale up linearly.
Paulos also points out how some of the numbers reported in the press are there only because we crave numbers. He's delightfully skeptical about economic forecasts, for example, which he characterizes as generally less sophisticated than football play-by-play broadcasting. And he takes on the "appropriately named" Laffer curve, the linchpin of Reaganomics.
I guess what I most enjoyed (besides the discussion of "the Jeffersonian model of many parallel processors" versus "the Stalinist model of one central processor") was his nailing of one of my pet peeves: The news story that begins, "This is not a scientific poll, but..." and then goes on to treat the results of their 900-number telephone survey as if they meant something.
Building an Internet library? Here are a few of the books on my Internet shelf. I'll skip all the intro books, the "Free Stuff" series from Coriolis, the personal accounts of lives wasted cruising the online equivalents of singles bars, and almost anything more than two years old.
HTML for Fun and Profit, by Mary E.S. Morris (Sunsoft Press, 1995), covers CGI (common gateway interface) scripting as well as the basics of HTML (hypertext markup language, the format in which WWW documents are written). The World Wide Web Unleashed, by John December and Neil Randall (Sams, 1994), and The Mosaic Handbook (O'Reilly & Associates, 1994) are two more good books if you write HTML. The O'Reilly book is an excellent introduction to Mosaic, the WWW, and HTML; the fat Sams book covers that ground and gets into Web-page planning and design concerns.
Connecting to the Internet, by Susan Estrada (O'Reilly & Associates, 1993), is about running wires and running the numbers. It provides checklists for getting on the net: projecting costs, performance needs, and so on. Internet Mailing Lists, edited by Edward T.L. Hardie and Vivian Neou (Prentice Hall, 1994), is useful if you want to set up a mailing list. Many people considering setting up Web pages might be better served by a mailing list. The Whole Internet User's Guide and Catalog, by Ed Krol (O'Reilly & Associates, 1992) is dated but authoritative, and a genuine classic. It's the book from which the other eight zillion Internet books steal, so why not go to the source?
Consider Netiquette, by Virginia Shea (Albion Books, 1994). You already know this stuff, but it's nice to refer to an authority when educating the clueless newbie. And Shea understands the most important law of the net: There is no law; these rules of netiquette are only convenient conventions.
I know I said I'd skip the intro books, but Christmas is coming up. You probably can't do better for your net-starved friends and relatives than one of the Internet for Dummies books (IDG Books, 1994), or for the Mac, Adam Engst's Internet Starter Kit (Hayden, 1993).
I've been mucking about with AppleScript, Apple's system-level scripting technology, since it was released. A new group of people is getting into AppleScripting these days: folks who are finding that Apple, hard as it is to believe, offers them the best platform for a Web server. Best security, best price/performance. Apple isn't used to bragging about its security or prices and hasn't done much of a job of getting the word out yet.
AppleScript is also very useful for managing a Mac-based server. That could be good news for two companies that have tied their fortunes to that of AppleScript--Userland and Software Designs Unlimited.
The introduction of AppleScript was fraught with promise, but Apple soon figured out that all the promise hung on users actually writing scripts. Trouble was, AppleScript suffered from invisibility: The script editor was, shall we say, modest; much of its vocabulary lay hidden in third-party applications, and the fine scripts that could serve as examples often ran invisibly in the background. In order to put a face on this technology, Apple wisely decided to bundle FaceSpan (from Software Designs Unlimited, Chapel Hill, NC) with AppleScript. FaceSpan is: 1. a front end for AppleScript that makes script writing easier, and 2. a tool for putting front ends on AppleScript scripts. Any script can become a stand-alone application with all the expected user-interface elements and all for about the same learning investment required to learn HyperTalk. Now FaceSpan has been nativized for PowerPC, given some powerful new features, and unbundled.
Is FaceSpan good enough to prosper from any AppleScript interest it helped generate? I don't know, but FaceSpan, plus AppleScript, probably represents the most rapid application-development system on the Mac. Faster, certainly, than any Windows-based system.
Meanwhile, as reported here last month, Userland has chosen to go the other direction: After struggling to keep a commercial AppleScript-related product alive (more to the point, a commercial AppleScript-competitor product), Userland has turned Frontier into freeware. Meanwhile, the company is moving rapidly into script-based tools for Web publishing; check out http://www.hotwired.com/Signal/DaveNet/.
In case you were wondering, the best source on botanical Latin is Botanical Latin, by William T. Stearn (Timber Press, Portland, OR).
In the Middle Ages, Latin was the universal language of intellectual discourse. Traveling scholars could be understood by local scholars because they all spoke classical Latin, a living language in which one could discuss anything from dinner plans to botany.
In 1690, philosopher John Locke laid down the requirements for a language that could support scientific discourse. These included the radical notion that a word ought to mean one specific thing. Linnaeus adhered to these rules when constructing his nomenclature of nature the following year. Before Linnaeus, a botanist would describe plants in language suitable for a chatty letter to his maiden aunt. After Linnaeus, plants were described in a highly formalized language in which verbs are largely eliminated and even the typography is formalized.
Botanical Latin, designed specifically for taxonomic purposes, was the result of Linnaeus's effort. From this beginning, botanical Latin evolved into a predictably extensible system of nomenclature that could serve as a language of this natural science. Arguably, it has also become as artificial and as formal a language as C or Pascal.
It even has a kind of ISO standard. In 1737, Linnaeus published what have come to be known as the Linnaean Canons. Example: "Generic names ending -oides are to be banished from the domain of botany." Or: "If we would not be considered utter barbarians, let us not invent names which cannot be derived from some root or other." The more recent International Code of Botanical Nomenclature gives detailed rules for producing new words based on the names of persons, such as, "When the name ends in a consonant, the letters ii are added, except when the name ends in -er, when i is added."
"When no fitting and meaningful name for a new genus comes to mind," Stearn advises, it's acceptable to rearrange the letters of a closely related genus. Maingola from magnola, for example.
With rare exceptions such as XINU, programming-language and operating-system nomenclature hasn't seized on this innovative technique of neologistics--yet. We can but hope.
Copyright © 1995, Dr. Dobb's JournalA Mathematician Reads the Newspaper
A Columnist Watches Television
A Mathematician Counts to Ten
An Internet Bookshelf
Scripting the Web Server
Obscure Language of the Month