In response to:
The Computer in Your Future from the October 27, 1983 issue
To the Editors:
It is always a bracing experience to have one’s feet put to the coals by the Torquemada of computing. However, two pairs of feet must be charcoaled equally: Feigenbaum is not the principal author of The Fifth Generation; his name appears first because F precedes M in the alphabet. We share credit and blame equally for this book. Before we’re totally consumed by all those righteous flames, we hope The New York Review of Books will oblige us by publishing an open letter to our old pal, Joe Weizenbaum, who reviewed our book, The Fifth Generation, for your readers in your October 27 issue.
There you go again. While all that stuff may come as a surprise to readers of The New York Review, your friends in computer science are really getting bored. For a decade (and in almost the same words) we’ve been either your favorite charlatans or your favorite archfiends. Oh, the ennui—the ennui is ever so much more painful than the vilification, to which we are by now well-adapted. It would be refreshing to be something else for a change. In any event, can we ever be redeemed? You bet we can. Read on:
At our request, all work on the Geriatric Robot has tottered to a halt to make way for a crash project: providing you with an expert system that will help you tell the difference between an obvious sendup and the real thing. You seem to have understood the rest of our book with equal discrimination.
But we sure like your substitute-this-for-that game. How about substituting “reading and writing” for every passage you quote about that other technology? Readers of this periodical ought to know how much of the world’s wickedness is supported by the written word—nay, only made possible by it—and then make a rational decision about whether they want to participate in such a mixed blessing.
Meanwhile, we are trying not to let it go to our heads that two mere authors were compared with so many powerful villains—Goebbels! Stalin!—oh, a tad dated, one could complain, given the share of villainy the globe has seen in these past three decades, but it’s the thought that counts.
Ever yours, even as you are ever ours,
Ed and Pamela
Thanks, editors, for being postman.
Edward A. Feigenbaum
Joseph Weizenbaum replies:
I feel compelled to say right at the start that Feigenbaum and McCorduck’s letter so offends my sense of what constitutes genuine human discourse, even in head-on clashes of ideas, that I have to overcome some embarrassment in order to respond at all. I do, however, believe it to be important to keep the record from becoming muddied.
Readers of this journal should not be led to believe that my critical review of these authors’ book is a continuation of a longstanding quarrel. Edward Feigenbaum and I have been friends for more than two decades—during which time, by the way, I wrote only highly complimentary words about both his early and his more current work in artificial intelligence.* Before the appearance of The Fifth Generation we had never to my knowledge quarreled over anything. I hardly know Pamela McCorduck at all. She interviewed me once for a book she was writing. The only other time I spoke with her was by telephone about four years ago. I dare say I wouldn’t recognize her were we to meet on the street. So much for “old pals.”
No, the review was sufficiently merited by the book; no other factors were needed.
One difference between an optimist and a pessimist is revealed in that, when the optimist elatedly announces, “This is the best of all possible worlds,” the pessimist responds, “That’s right.” I have lately often had the feeling that I am living this little joke. The partisans of technology among whom I, as a member of the MIT community, live, enthusiastically paint pictures of worlds filled with products of their genius. They promise they can actually create these sources both of their optimism and their pride. When I know enough of their fields to have to believe them, my depression sometimes deepens. Perhaps I am more “optimistic” than Feigenbaum and McCorduck; I think that even the machinery they invent as only a “sendup” can (and will) be created and, what is more important, that it fits naturally into the world the rest of their book describes. As I see this world as having its horrifying aspects, my despair grows. I must say that their explanation that they intended the geriatric robot to be a joke doesn’t help. Perhaps the whole book is a joke.
These remarks exhaust the content of Feigenbaum and McCorduck’s note. But because my review of their book called their scholarship into question (I charged them with supporting arguments by reference to nonexisting documents and with consistently failing to give bibliographic citations whenever they alluded to views contrary to their own) and because it cast doubt on their authority as scientists (no scientist would believe that computers produce information, for example), I would have expected some attempt on their part to shore up their reputations.
As a minimum, commonly accepted rules of scholarship require them to identify the “work” in which, they say, Von Neumann recorded his “long argument that computers would never exhibit intelligence.” It would be good to know how and by whom that argument was finally overcome. And, by the way, into which of the four categories the authors claim all arguments against artificial intelligence fall does Von Neumann’s fit? Beyond that, I am curious to see how Feigenbaum and McCorduck would defend their assertion that computers produce information, indeed that therein lies the “essence of the computer revolution.” They might characterize that bit of nonsense as merely a “sendup” and the missing citations as exercises deliberately left for the reader. Then again, perhaps these and other faults of the book are simply indefensible. That, certainly, is my view.
Computer Power and Human Reason (W.H. Freeman, 1976). ↩