Indifferent Designers, Unuseable Designs
May 14, 1998
Book Review
Charles M. Strauss
Machine Beauty. By David Gelernter, Basic Books, New York, 1998, 166 pages, $21.00
This book infuriated me! Not that being infuriated is necessarily a bad thing, you understand. A very pleasant part of the experience of higher education is argumentation, one-on-one discussions or multipart bull sessions, arising from the interplay of young and active minds. David Gelernter raises all sorts of important points and asks all sorts of interesting questions. But I find his focus and continuity weak to the point of invisibility, and his own answers to his interesting questions seem to me either distressingly incomplete or, even more distressingly, just plain wrong.
Let me state my main bias--I am, by profession, a grunt programmer, one of those guys right down there in the trenches actually writing code in real languages to run on real machines. When academicians start telling me "how great it is going to be" (as the old IBM salesman joke has it), my hackles rise. My peers and I have heard that siren song far too often over the years, especially from the artificial intelligence community, to listen with anything but annoyance. Whatever became of the 5th Generation project (remember that one?), anyhow?
One detail from the book illuminates the difference in outlook between the author and this reviewer. In two places Gelernter writes of the tendency of flat roofs to leak as a regrettable by-product of certain aesthetics in architecture (of the Bauhaus school, on page 13, and of Frank Lloyd Wright, on page 17). Well, my family was in the roofing business, and I can assure Gelernter, as well as all the architects in the world and all the unhappy people dwelling under leaky flat roofs, that flat roofs do not have to leak. The problem is rather that architects do not listen when roofers tell them that the buildings they have designed are going to leak, nor do they respond to suggestions for correcting their designs so that the roofs will not leak. This problem is systemic rather than technical; i.e., it is a result of a flaw in the system for designing buildings and implementing those designs in our culture.
No member of SIAM can possibly be indifferent to the beauty and power of a great notation system: the d/dx of calculus, the matrix representation of a linear operator in n-space, the commutative diagrams of algebraic topology. The sheer amount of work and thought encapsulated in these notation systems excites our gratitude and appreciation. But the beauty and the power of the notation arise from the confluence of economy and results, not from economy or results alone--and our aesthetic standard, by which we judge the notation so elegant, is not the only possible standard. We, as professional mathematicians, know all about metrics and norms. We also know that the L� norm is useful under some circumstances and the SUP norm is useful under other, possibly quite different circumstances. So does the rest of the world, just not in such exact terms.
One of the problems of Machine Beauty is that Gelernter attributes problems and remedies to natural causes, e.g., a deep-seated human or cultural dislike of elegant design. On page 9 we read, "At base, machine beauty rubs us wrong." The main point of the first half of the book is that the PC with Microsoft's operating systems beat out the Macintosh because in the public's mind Elegance = Cute = Frivolous = Unsaleable. This analysis is perfectly correct, and the point is one that I had not really focused on, so I am glad to have had this aspect of the Macintosh's relatively weak showing pointed out.
In hindsight, though, just how surprising is this discovery? Who in the 1980s had the most money to spend on computers? Managers in businesses, right? And what do these people care about? Control, for one thing; uniformity, for another; oh, yes, and making money. We all read "Dilbert," don't we? But it's not that these people go out of their way to hate elegance--they totally ignore it. The difference between active dislike and blank incomprehension is important, and the difference in the remedy for these two states is important, too.
The indifference of designers in our society to the useability of their designs (forgetting for a minute about the poor programmers, who have enough to answer for without being trashed for bad user interface design) is a matter of common scandal. As long as Donald Norman (in, for example, The Psychology of Everyday Things) can easily cite common and yet mind-bending examples of poorly designed doors, kitchen stove and refrigerator controls, light switches, etc., ad infinitum, our civilization obviously has a still-open problem to confront and solve.
From a systemic view, part of the answer to the problems of the Macintosh lay in (gasp) better advertising. Jerry della Femina points out that good advertising teaches things about the product that the audience does not already know. Does the phrase "impedance mismatch" ring a bell? A product too far advanced for the market is just as bad for sales as an out-of-date product. (I personally attribute a lot of the consumer preference for the PC over the Mac to what I remember as the roughly 50% additional cost of a Macintosh, as mentioned by Gelernter on page 38.) Even Jim Bouton's manager, Joe Schultz, knew that "well, it's a round bat and a round ball, and you've got to hit it square!"
The second half of Machine Beauty discusses some possible ways to extend the desktop metaphor, which Gelernter rightly views as a way-station on the proper road to the effective use of computers rather than the final "best" model of human-machine interaction. His examples, however, fill me with horror. His "Lifestreams" example (beginning on page 102), intended to solve the "where is that piece of paper?" problem, reads like one of the dystopian science fiction stories that I remember from the Groff Conklin anthologies of the 1950s. The idea of a palpable stream, suspended in time, of all the documents I ever created, read, or handled, terrifies me. How would you like to see your entire intellectual history displayed (or at least displayable) in this way?
We see at work here what David M. Balabanian calls the thinking of the "newly smart," and the result of this sort of thinking is what we creatures of the 1990s have learned to call "the law of unintended consequences." Gelernter, like all of us, has continually run up against Zipf's Law or its corollary, the 80-20 rule. He has helped to devise a mitigating technology but has apparently given little thought to any possible downside to this technology. The smart money rides on the existence of a downside. The telephone is a great invention, but telemarketers are a plague on the land. The cellular phone frees us from the anchor of the office but at the same time enslaves us even in what used to be our free moments. Dwight Eisenhower caused a great interstate highway system to be constructed, but interacting with American civilization, where the urge to "light out for the territory" is embedded in the national psyche, these highways have given rise to a huge abandonment of the nation's cities. Let's start thinking about all the possible consequences, not just the pleasant or immediate consequences, of our grand schemes.
But enough of my reactions. You might well want to give this book a try and see if it provokes the same sort of instinctive disagreements in your soul.
Charles M. Strauss is a senior member of the technical staff at the Charles Stark Draper Laboratory in Cambridge, Massachusetts.
