8 May 1993

Fuzzy, Inc.

FUZZY THINKING: The New Science of Fuzzy Logic. By Bart Kosko. Hyperion, $24.95.

FUZZY LOGIC: The Discovery of a Revolutionary Computer Technology--And How It Is Changing Our World. By Daniel McNeill and Paul Freiberger. Simon & Schuster, $22.

AI: The Tumultuous History of the Search for Artificial Intelligence. By Daniel Crevier. Basic Books, $27.50.

GAME OVER: How Nintendo Zapped an American Industry, Captured Your Dollars, and Enslaved Your Children. By David Sheff. Random House, $25.

Like most interesting things, the emergence of "fuzzy logic"--the most highly hyped trend in pop science to come down the pike since chaos theory--is attended by plenty of good, old-fashioned infighting, backbiting, and petty bickering among its advocates and detractors. As controversial as it is seemingly innocuous, fuzzy logic tempers the boring certainties of Aristotelian logic with a more refined appreciation of vagueness, paradox, "graded memberships," and matters of degree.

Dubbed the "St. Paul" of fuzziness by a Los Angeles Times writer, Fuzzy Thinking author Bart Kosko is a 33-year-old assistant professor of electrical engineering at the University of Southern California and a disciple of Lotfi Zadeh, who introduced the notion of FUZZY SETS in a 1965 paper drawing upon the work of Max Black. As Daniel McNeill and Paul Freiberger describe it in their less self-serving Fuzzy Logic : "Suppose two people in a living room are watching Bonfire of the Vanities on the VCR. The (fuzzy) set of annoyed people in the room is Sam/0.85 and Pam/0.80. Its complement, the set of not-annoyed people, is Sam/0.15 and Pam/0.20." (37) Fuzzy logic deconstructs the boundary separating inclusion and exclusion by assigning memberships in both set A and set non-A, but in degrees. (And if you're glad THAT's out of the way, be forewarned that both these books deal with the permutations of set juggling extensively, if not technically.) The implications for engineering, manufacturing, and consumer electronics are both more and less than meet the eye.

Whether or not fuzzy logic rates as a "new science," as the subtitle of Kosko's fuzzy manifesto would have it, depends on how deeply you buy into both its personality cults and marketing fantasies. While nearly everyone characterizes Zadeh as a sweet old guy, Kosko comes off as a theorem-crunching Bruce Lee. A composer, karate black belt, science-fiction writer, and would-be movie star along the lines of Stephen Hawking, the young zealot grocked his personal fuzzy revelation--the Subsethood Theorem--during "an empty zazen, the best kind" (58), the kind men like. Where other mathematicians dis fuzzy logic as "the cocaine of science" and as "a kind of scientific permissiveness," (Kosko, 3), this one-time Strategic Defense Initiative consultant sees in it nothing less than the Buddha's victory over the Aristotelian tradition.

For as all of these books constantly remind us, our nation is engaged in a grim struggle with the Japanese economy's two-pronged offensive, combining protectionism with technological . God forbid that Japan should beat us to the fuzzy punch. Though they have! They have! As you read this (and if you have trouble sleeping tonight, I'll understand), Kosko claims that no fewer than 10,000 fuzzy theorists are slaving away in China (67). While our children sit mesmerized in front of their Nintendo Entertainment Systems, Japanese superscientists are presumably perfecting the "smart cars," novel-writing computers, "molecule-sized soldiers of health" (McNeill, 13), and "maybe even sex cyborgs modeled after the pop and other stars of the day" (Kosko, 264) Kosko predicts fuzzy science will eventually provide (he has signed up for a cryogenic freeze in anticipation). Until then, however, we must settle for the more prosaic products now available; these include washing machines, vacuum cleaners, elevators, and microwave ovens that can intuit more precisely the needs of us the consumers.

"Puffery" has contaminated the field," according to McNeill and Freiberger, who make a much better case than Kosko for how fuzziness could help humanize technology. Noting that the fuzz buzz has informed the 20th century since Nietzsche, they point out in passing how science, art, politics, and the humanities--from Robert Wilson to the European Economic Community--have been dissolving boundaries and shrugging off tautological reasoning with a vengeance for nearly a century. Allowing Kosko to role-play the upstart paradigm shifter and go out on his technoporn limb, they emphasize fuzziness's inherent simplicity, concluding that fuzz "seems unusually well-suited for American firms, especially those facing a shortage of capital and a decline in consumer confidence compared to their Japanese rivals" (282). Who needs NAFTA?

Information wants to be fuzzy; as complexity increases, so does indeterminacy. So far, this has been the biggest drawback with what Daniel Crevier terms quite accurately the "tumultuous history" of artificial intelligence in his cotton-mouthed and acronym-leavened historical survey, AI . Like the fuzzballs, Crevier forecasts spectacular cosmic results from the minute linguistic and cognitive particularities the "artificial intelligentsia" (115) struggles mightily to accommodate. Where fuzzy logic seeks to model systems in words rather than mathematical symbols, AI attempts to accommodate computers, common sense, and natural language. The bad news, however, is that "You cannot define even the most innocuous and specialized aspect of human usage without reference to the whole of human culture" (108).

Unlike fuzzy logic, which bubbles quietly in the margins of mainstream mathematics and probability theory, AI takes a rollercoaster ride in the halls of academe, especially MIT. The AI community is divided into starched-collar "neats," such as John McCarthy and Alan Robinson, and rumpled "scruffies," whose ranks include MIT's Marvin Minsky and Seymour Papert. The former, obsessed with logic and syntax, stake their theoretical turf in opposition to thinkers more willing to accommodate semantic imprecision (161, 183). Guess where fuzzy logic fits in.

In some sense the virtually intelligent computer may always remain alien to science while, in the words of William Gibson, "the street finds uses for things" that we can't yet imagine. Chip beats brain when it comes to computational speed, and while computers today play grandmaster-level chess, they remain little more a "void parody" (198) (Crevier's words) of human thought in the software department--brittle and unreliable "solution suppliers" (205) for the most part. Crevier foresees AI futures ranging from a malevolent computer "Colossus" (borrowed from novelist D. F. Jones) to dehumanizing control mechanisms to expert systems that will (stop me if you've heard this before) liberate us from work, equalize wealth, clarify our humanity, and perhaps leave us in the dust when machines deem it time to colonize the universe. Speculative ethnobotanist Terence McKenna, however, has suggested that machines consisting of earthly metal, plastic, and silicone perhaps represent nothing less than the revenge of Gaia.

None of these possible futures, however, seem quite so pretty as that imagined by Tom Maddox in his 1991 novel Halo , in which an enriched version of Apple's Newton serves as an elite technopal on the verge of self-awareness in "machine space."

He and the memex had worked together for more than a decade, the memex serving as confidante, advisor, doctor, lawyer, factotum, personal secretary, amanuensis, seeing him in all his moods, taking the measure of his strengths and weaknesses, sharing his suffering and joy. And he thought how honest, loyal, thoughtful, patient, kind and . . . SELFLESS the memex had been--inhumanly so, by definition, the machine as ultimate Boy Scout; but one, it turned out, with complexities and needs of its own. (106)

There are those, of course, who feel they have already seen the future of artificial intelligence and it looks like an incredibly seductive videogame. Nintendo has been Japan's most successful company since 1988, providing a product that merges game and players in an intense cognitive, emotional, and physical (as in pain-killing and endorphin-generating) relationship that transcends more pedestrian variations of cyberspace. Written with the cooperation of its subjects, David Sheff's overly admiring Game Over charts the company from its 1889 origin as a playing-card manufacturer to its present world domination. And the apparent dispassion and dry-ice business acumen with which that end was attained would make Bart Kosko's zazen revery wither in the void.

Nintendo's future lies in the complexly networked expert systems imagined in fuzzy-logic and AI researchers' dreams. Linking the 33 percent of American homes in which the system is present (as compared to a 15-percent computer saturation), Nintendo plans a network that would ultimately "dwarf Prodigy and its competitors" (390). The company's low-end PCs would do all those things computer nets promise to do, and then some. "'We do not see borders in this business,'" blustered Nintendo potentate Hiroshi Yamauchi some years ago. "'Some countries may be too poor or have heavy tariffs on imports, but with those exceptions we will go anywhere in the world. There are no borders.'" (415)

Borders, of course, are precisely what information technology redefines and erases. "Information wants to be free" goes the hacker's mantra, offsetting technophobia with anarchic idealism. The mind of the machine will have its own agenda, of course. And as the media multiply, if not replicate, it seems only natural that their complexity will increase as well. Don't worry. The kids are alright, and will adapt to whatever environment Nintendo markets their way so long as it doesn't pose too grave a threat to our anthropomorphic intelligence. Be it ever so fuzzy, the garbage-in, garbage-out rule will still apply.