The Future Does Not Compute

This is Chapter 3 of The Future Does Not Compute: Transcending the Machines in Our Midst, by Stephen L. Talbott. Copyright 1995 O'Reilly & Associates. All rights reserved. You may freely redistribute this chapter in its entirety for noncommercial purposes. For information about the author's online newsletter, NETFUTURE: Technology and Human Responsibility, see http://www.netfuture.org/.

Computers are tools of the past. They are perfectly designed to aid our understanding precisely insofar as it is a past-understanding. For example, if we want to know when lunar eclipses will occur in the year 2053, there's no better tool for figuring it out than a computer.

“But wait a minute. 2053 isn't in the past. That's a prediction of the future.”

Well, yes, in a trivial sense. More reasonably, you might say it's the past simply projected into the future. And the projection involves nothing new; it's nothing but the past. Everything in the “prediction” was already fully implicit in the celestial configuration of, say, 1857. All we're saying is, “Here's what the past looks like when projected onto the year 2053.”

What happens when we really turn our computers toward the future -- or try to? All too often, disaster. Disaster, for example, in the form of decision support systems wrongly applied. Human decisions clearly are (or ought to be) matters of the future. We make decisions in order to choose a future different from the one now approaching us. No analysis of what is, and no set of previously calculated questions or heuristics, can remove the burden of choice. When I choose a future, shall I reject 1% of the analysis -- which is to say, 1% of the past -- or 99%? What is to guide me -- more analysis of the past?

How shall I choose my wife?

Ask yourself about the critical decisions in your life. How did you choose your present vocation? There are, of course, very sober ways for doing this. Begin by taking a few psychological tests to inventory your skills and aptitudes, personality type, likes and dislikes, character, stability, financial requirements, geographic preferences, physical characteristics (strength, endurance, handicaps), and so on. Then align this inventory with a similar analysis of all the possible vocations (basic skills required, profiles of the most successful people in each vocation ... ). Finally, identify the closest fit and (presto!) there's your future -- presumably a well-adjusted, profitable, and happy one.

But, no, something's not right here. Have you, or has anyone you know, ever made an important decision by weighing all related factors, adding them up, and then obeying the sum? This is not really your future we're talking about; it's your past. The real question is, what do you choose to become -- despite what you are now? What future not already embodied in the past will you embrace? Of the great figures of history, where would they be if they had merely hewed to a reasonable future? Joan of Arc. The first black in a white Mississippi college. The first woman doctor. The soldier who dives on a hand grenade to protect his comrades -- what sort of a future is that? Yet we honor him.

The psychologist, Alfred Adler, went so far as to make a rule of the fact that outstanding people typically work against their profiles and their past. /1/ He tells us of various painters who had eye problems, and claims that 70% of art students were “found to suffer from some optical anomalies.” Musicians, he notes, have often suffered from ear afflictions -- leading to deafness in the case of Beethoven and Robert Franz. Clara Schumann reported hearing and speech difficulties in childhood. And then there's Demosthenes, the stutterer, who became the greatest orator of Greece. What sort of decision support system might have been useful -- or destructive -- to these people?

Or take marriage. Shall I choose a wife reasonably, because all the indicators point to our being well-adjusted and happy, or shall I plunge into a future I cannot fully see, but that I am strangely, mysteriously, drawn to, dimly recognizing something of myself (but not yet myself) in my partner? Is there really a choice to be made between the perfectly compatible marriage of the inventory-takers and the reality cited by Adolf Gueggenbuhl-Craig? Marriage, he says, is

a special path for discovering the soul .... One of the essential features of this soteriological pathway is the absence of avenues for escape. Just as the saintly hermits cannot evade themselves, so the married persons cannot avoid their partners. In this partially uplifting, partially tormenting evasionlessness lies the specific character of this path. /2/

Surely we may as well accept this from the start (and you can see such acceptance in every good marriage), for that is the way it will turn out in any case. Can I gain anything truly worthwhile in life without suffering (and discovering) the unexpected? But how shall I program into my computer the unexpected, or the equation of suffering and reward?

Every question about the future -- every human question -- is like this. We strike out into the unknown, with a hope and a vision perhaps, but without an adequate “basis” for our decisions. After all, a perfectly adequate basis would mean the decision was trivial, because divorced from questions of human destiny. Unfortunately, however, broad areas of our lives have fallen under the spell of the computational approach, where we imagine the computer -- the past -- to hold the secret of a future that is, therefore, no longer a future.

Computing to work

Look, for example, at business. In many contexts we take it for granted that our businesses should be managed as matters of the past. Analyze the market precisely enough, design the product to fit the analysis, and the bottom line is guaranteed -- because it is already implicit in the past. At least, that is the ideal of “perfect analysis and planning” toward which we strive, and we measure our success against it.

What has fallen out of this picture? Just the entire meaning of work, just the whole human reason why people band together and direct their creative energies toward productive ends. Ends -- things worth pursuing. Goods that are good, services that serve. The corporation has divorced itself from questions like “What task do I wish to take up in the world?” and “What is the future we wish to create?” We form our companies only to learn that they no longer embody our ends, but somehow have a neat, predictable logic of their own. We choose a future only to find it taken out of our hands, exchanged for a permutation of the past. /3/

The computer, one might almost say, was invented as an inevitable refinement of the corporation. Much of the early work in artificial intelligence came out of schools of management, and there was a great deal of excitement -- this was back in the 1960s -- about how computers would soon take over all the business manager's hard decisions. And make a better job of it. Computers didn't bring anything essentially new; they were just going to be better machines than we had yet managed to be.

Weighing the past is critically important, in business as elsewhere. It helps us to see the current playing field, identify constraints, compare options against an established framework. For this, computers are immensely valuable. But they do us no good if, in the process, we lose the future -- if our thinking becomes so shaped to a knowledge of the past that our courage fails us when it comes time to break the machine and declare for the unjustifiable.

When judges are computers

There is a strong case to be made -- although I have not made it here -- that our acquired skill in analyzing the past has largely hidden the future from us. This is hardly surprising when you consider that our past-knowledge is commonly held (at least in academic and scientific circles) to imply that there is no future -- no future of possibility, no future not already determined by the past.

The horror of such a view -- once it links up with the computer's ability to preserve the past as our future -- echoes through a query by one of the pioneers of artificial intelligence. John McCarthy asked, “What do judges know that we cannot tell a computer? /4/ The question is grotesque. Some theorists may, after all, have succeeded in reducing their own tasks very substantially to mere computation. But a judge must forever be asking: by what metaphor can I understand this person before me? How does he differ from anyone I have seen before? How may I grasp the difference, and what decision may I impose, consistent with law, that will give him the best chance of coming to himself?

There remains one convenient thing about an impoverished, ossified future: it computes. Of course it computes -- it is exactly what the computer was designed to give us. Our true future, on the other hand, can never be computed -- so long, that is, as we retain the courage to call it into being at all.

References

1. Adler, 1964: 21-43.

2. Gueggenbuhl-Craig, 1971: 41.

3. Chapter 8, “Things That Run by Themselves,” explores these necessities in considerable detail.

4. Quoted in Weizenbaum, 1976: 207.