The Wisdom of Crocodiles

by Paul Hoffman

  Winnicott brightened. ‘Well, I’ve come across a few monsters in my time but I’ve never anything that claimed to be the most difficult. What is it?’

  ‘El3,’ said McCarthy, wondering if Winnicott knew this as some shibboleth of real crossword puzzlers – old-hat stuff. But from his expression it was clear he had never heard of it.

  ‘A postal district?’

  ‘Oh, no. At least, not on the face of it. It’s just an E, with thirteen letters in the answer, whatever it is. Of course it could still be something to do with a postal district. Anyway, I thought you might enjoy it.’

  ‘Oh, absolutely. Though I probably won’t be able to sleep till I solve it. Does your brother know the answer?’

  ‘Yes, but he says he won’t tell me till he comes home next Christmas.’

  What Winnicott wanted to do was get out a pen and master it like some lexicographical lion tamer, but he felt this would be rude. Conversation was being made with him and he was obliged to return the favour. ‘Do you think this new computer is going to be useful?’

  ‘It might be. It’s just a field trial, but if it does half of what they’re hoping we should definitely be involved. It won’t cost us anything, so on the face of it we’ve got nothing to lose and quite a lot to gain, maybe even in the short term.’

  Ten minutes later Anne Levels waited to begin her lecture in the Troilus and Criseyde Media Centre, feeling more like a second-hand-car dealer than a scientist. This was a sales pitch as much as a lecture, and Anne was the true daughter of devout shop-owning parents in whom a genuine piety, a profoundly held faith in the notion of caveat emptor and the deep beauty of a balance sheet in the black had no need of reconciliation. Whether it was an idea or a computer program, she loved to sell.

  Anne smiled her most winning smile. As a young woman she had become notorious for the devastating trick she had pulled in America on an audience filled with the most eminent men in the field of artificial intelligence. At the time she had felt only youthful delight at being the one to smash a deeply held orthodoxy, but over the years she had often regretted that she had made a fool of so many influential people in such a public and humiliating fashion. In the years since she had upset so many at her show-off debut she had learnt to be more careful. She was now ready to stroke those who needed it. Some of those present were academics. There were the pure scientists, as unsullied in their vocation as any nun, and the age-defining business dons, Mekons of money, all technology parks and BMWs. Many of them loathed her. The third audience consisted of three Japanese executives from Hitachi, accompanied by a translator. Hitachi were paying for the trials that she was formally about to announce. There was also, unknown to her, someone else present: Steven Grlscz. Although she knew him only from his occasional articles in some of the less well-known medical journals, she was immediately drawn to the ideas she found there, essentially because they were like her own. She had sent him an invitation but he had not replied. A deeply worried man, he had only decided to come to Canterbury at the last minute as much as anything for the distraction.

  But today her peers were here as guests to witness the beginning of what she hoped would be a landmark occasion: the beginnings of a revolution, the biggest technological advance since the invention of the motor car. At the age of thirty-eight she was coming into bloom like some extraordinary flower, and not even the modesty of the way she dressed could prevent the impression she gave of being about to spill out of her clothes and surround you with a riot of white skin and curve and long black hair. She seemed to be growing as you watched, her hair lengthening, her lips opening; everything about her seemed full of juice, like a great warrior champion of the erotic. Since she was thirteen, Anne Levels had known how much she was desired, aware of lechery for her breasts and thighs and the turn of her throat. But she was about to set in motion the discovery of how deadly it can be to be loved for what you are: lusted after for your intelligence, craved for your under-standing; that there is lechery over sweetness of temper, greed for compassion. Lust for love.

  Within six months four of the people in the room would be dead. She stepped up to the podium and began to sell the ideas that, in time, would be responsible for three of those deaths.

  ‘I am delighted to announce that over the next few months NEMO will begin an extended field trial. Some of those involved have asked for the details to remain confidential but I can confirm that we will be working with the Department of Social Security, the Fraud Secretariat and the Department of Structural Engineering at Imperial College. It is, I must stress, merely the first phase of what will undoubtedly be a long process. In the short-term, nothing of great immediate practical value may emerge. The long-term benefits, if we succeed, may be incalculable. NEMO has the potential to become the most important technological innovation since the motor car.’

  Having startled the audience with this deliberately provocative claim, she went on to settle them down by drawing an apparently more familiar picture. Knowledge about the world, she said, had probably doubled between the birth of Christ and 1750. It had taken another hundred years for it to double again. Now it was doubling every five years. She realised, of course, that they were all aware it was becoming difficult in the light of all this to keep abreast of developments – even in their own fields. There was a stir of regretful acknowledgement. No one could disagree. Keeping up was hard to do. The greatest disaster in the history of knowledge’, she continued, ‘took place at the end of the third century AD, when most of what the world knew was destroyed in a single night when the great library of Alexandria burnt to the ground. We can scarcely imagine what the scholars of the city must have felt as they stood next to the ruins to bear witness to this dreadful loss. They must have felt that civilisation itself had come to an end in only a few hours, and all because of a simple act of carelessness.’ She looked around the room. ‘We are facing a similar apocalypse.’ The audience was satisfyingly stunned by this assertion.

  ‘We hear much about the information superhighway, with all that this implies in terms of coherent flow, of slipways, lanes, speed of exit and ease of distribution.’ She looked up and paused for dramatic effect. ‘But the highway is not delivering information, it is dumping it. Then it’s immediately dumping more on top of what it has already dumped, and then dumping more on top of that. In the future, knowledge will not be consumed by fire, as were the half a million scrolls in the great library of Alexandria. It will be smothered. And it will be smothered by itself. An increasing number of things will be understood, but we will not know that we understand them.’ She added regretfully, ‘Where, indeed, is the knowledge we have lost in information? Even as I speak, a scientist is discovering things that have already been discovered by someone else. Another could make a breakthrough if only he knew one piece of information that lies buried in a book next to the one he’s looking through, or in a memory bank unlooked at for want of the right question typed into the computer.’

  With the mention of the computer she had reached the next stage of her pitch, for she was most definitely selling them something, thought Grlscz, as he watched with increasing attention.

  She tore into the uselessness of computers. They were expensive, cumbersome, over-hyped, a huge drain on company resources, fickle, inefficient, absurdly difficult to use. ‘Despite the commercially hyperactive nature of the computer industry, recent research makes it clear that in parallel with the gains being made in commercial and corporate efficiency, huge amounts of money, probably in the thousands of millions, are being wasted trying to make computer systems deliver on promises which are often perilously close to fraudulence. And speaking of fraudulence, the final bill for the millennium bug has been calculated as approaching four hundred billion dollars. Ask yourselves what good the money wasted on this one failure alone could have done had it been spent on something useful.’ She paused to let this denunciation sink in, looking like some voluptuous Calvin of the digital age railing against the soft and hardware whores of Ba
bylon.

  ‘The problem, of course, is the software. Computers themselves are on direct line to ever greater speeds and memory capacities, all at a falling cost. But the means of exploiting these advances lags further and further behind, although this is often disguised by the pseudo-usefulness of many systems. The most pressing issue for those of us involved in artificial intelligence is how we can make our systems usable and effective rather than indulge ourselves in grandiose posturing about the innate humanness of our creations.’

  Grlscz saw that this barb was an act of deliberate provocation since several of the men in the audience looked as if they’d been forced to swallow bleach.

  The next stage of the pitch had been reached. She took a sip from a glass of water and then continued. ‘We have a saying in this country to describe the way someone combines two previously unconnected pieces of information to produce new knowledge: I just put two and two together. Adding two and two is a function that the cheapest kind of pocket calculator can now accomplish more quickly and accurately than the best mathematician. But putting two and two together is a product of desire, will, commitment, inventiveness, individual autonomy. It draws on a memory that has nothing in common with the banal storage of a computer. Our memory is a living, growing capacity that creates and forgets, then reconstructs the world in the shape of something entirely new. This is something forever lost to the world of the computer. But something forever lost to us is the ability to store anything like the vast amount of knowledge that there is about the world in the straightforward way that we need to process it and make it work for us. We are losing contact with what we know and unless we can create something to mimic that autonomy, mimic not the adding of two and two but the putting of two and two together, then the future of human knowledge is doomed to a terrible dysphasic inefficiency.’

  There was a ripple of discussion among the Japanese. One of them, the interpreter, raised his hand.

  ‘Ah . . . dysphasic, Dr Levels?’

  This threw her, not least because she could see O’Connor, the CEO of Machine Intelligence, looking at her darkly from the back of the auditorium. ‘It’s . . . ah . . .’ Her mind went blank. There was an embarrassing silence.

  ‘It’s a symptom of brain damage,’ said O’Connor coldly, ‘where the—’

  ‘Where there’s a damaged access to language,’ Anne interrupted in a desperate rush. ‘You have trouble finding words.’ She laughed weakly at herself. ‘You can’t get . . . access to . . . you can’t find the words to express complicated thoughts.’

  She looked at the translator hopefully but his expression was full of doubt as he turned to the others and spoke in Japanese. Whatever he was saying, his uncertainty required no translation. The room waited anxiously for them to finish, but when the man haltingly stopped, the other Japanese nodded self-critically and smiled, as if the blameworthiness of not knowing the meaning of dysphasic was profoundly mitigated by the sheer usefulness of the term now at their disposal.

  Anne started again, chastened and cursing the insomnia that had kept her up half the night. She knew she had to speak to a mixed audience and to have left in such a word was so stupid. She had missed an obvious problem because of tiredness. She knew the revisions should have been done earlier. Everything in her life would be all right if there were only three and a half per cent more time in the world.

  ‘With this problem in mind, I began in 1989 to observe people in the act of creating hypotheses. This was unhelpful at first because I concentrated on my colleagues, assuming, reasonably enough, that being expert hypothesisers they would repay the closest study. But it was only when a young family moved into the house next to my flat, which overlooked their garden, that I started to see where I was going wrong. I began to watch their three children, and it was then I started to realise there was something fundamental to creating new knowledge that was too readily hidden by experts in such creativity. That something was the capacity to make frequent implausible errors. My neighbour’s son was particularly helpful. At the age of six he is old enough to perform complicated actions skilfully but not old enough to have devised too many abstract mental models of a kind that would prevent him from carrying out his more eccentric hypotheses. On one occasion he spent nearly an afternoon trying to work out a way of knocking an apple from the tree at the bottom of the garden. He tried standing on a chair, poking it with long sticks, but when he finally started to try and knock it down by throwing a sharp knife on the end of a string, I was obliged to put my role as a good neighbour ahead of that of disinterested scientist and shout at him to stop.’

  The audience chuckled indulgently at this.

  ‘On another occasion when I was looking after him in his own house for a couple of hours I found him scraping a piece of burnt toast with a screwdriver while wearing a pair of swimming goggles. When I asked him what he was doing, he explained that the goggles were to prevent the dust from his scraping of the burnt bits of the toast from getting in his eyes. In watching him arrive at a solution to the problems that faced him, I asked myself how he had arrived, not at the correct ones, but at the incorrect ones. He was making guesses about the world and how it could be manipulated in line with his wishes. But how many, I wondered, before he arrived at the one he decided to act on. How are they generated and, if hypotheses are rejected before testing, on what basis are they rejected? Maybe we will find this out, but not in my lifetime and not, I suspect, in any other. I have become convinced that the nature of human creativity lies in its relationship to error and instability, its endless capacity for making wrong hypotheses that are not wild guesses but which may be very close to that. Like fractal mathematics, it is possible that the human mind takes shape at the boundary between chaos and order. This is the essence of our success as creatures: we make mistakes all the time but still manage to survive and prosper. But I felt defeated by the unfathomable nature of the way in which human beings make mistakes in useful ways . . . guesses, intuitions, idle dreams, the putting together of goggles and toast with a screwdriver.’ She gestured with her left hand in a circling movement as if to indicate the airy ungraspability of such things.

  ‘So I tried putting two and two together myself. Revealingly enough, one of the most important insights in the development of NEMO came about through pure luck. I picked up a magazine left in the waiting room of the Accident and Emergency department of my local hospital when I was waiting for a friend who was having an X-ray after falling over and banging his head. The magazine was the International Journal of Aeronautics. Goodness knows what it was doing there – it was at least two years old. One article concerned a radical new approach to fighter-plane design. Apparently the problem with building such planes is that, conventionally, they are designed to be aerodynamically stable. It turns out that this very stability is the problem. It means that a great deal of effort has to go into making them change course, something extremely important for a war plane as it tries to dodge attacks by rockets and other planes. Manoeuvrability is of the essence. Someone wondered what would happen if you made a plane inherently aerodynamically unstable but used computers, monitoring and adjusting continuously, to keep it stable during normal flight. The advantage of the plane being inherently unstable is that when it’s attacked, much less effort is needed to get it to turn and twist its way out of trouble. I began to wonder if the human brain works in this way. Why is the brain so much unlike a computer? It jumps here, there and everywhere. Its immense complexity seems to thrive on being able to make the kind of leap that aerodynamically stable computers find impossible. My guess is that we have a system in the brain for keeping it stable for ordinary everyday functions and for work that requires the intensely methodical – but that its extraordinary creativity is a by-product of a fundamental instability. In the human brain, evolution has created a mixture of the reptile, mammal and human, which combines maximum instability with maximum order to create the most effective result. This must be a staggeringly robust alliance given the extre
me contrasts involved. But it would explain why there is so much contradiction in human nature, and why the claims that computer software is a model for the human mind are such nonsense. Natural selection is brutal. If we did not need the non-rational, the emotional if you like, we wouldn’t have it. Rationality is only one part of understanding, and I include scientific understanding.’

  She looked around the room.

  ‘If the supreme triumph of reason is to recognise its limitations, then we must recognise that it is illogical to believe that the non-rational is the same as the irrational. Whatever they are, emotions are not a form of bad reasoning. Intuition may be a name for the program that mediates between reason and emotion, but I can’t even think what form such a program might take or even if talking about it in such terms makes sense. I doubt we will ever convert emotion into knowledge.’ She paused for a moment.

  The next stage of the hard sell, thought Grlscz.

  ‘Fortunately I am convinced that we don’t need to engage in such a potentially fruitless search. I am going to suggest that a way out of the problem of too much knowledge will not elude us. It occurred to me that we have another model for the generation of highly complex and useful errors: evolution itself. Evolution is a mechanism where error happens creatively – inefficient errors are eliminated, useful ones survive and prosper. But this is the product of a blind mechanism that works every second of every hour across vast stretches of time without pity or remorse. There is no respite, neither good nor bad intentions, no desire – there are no intuitions or dreams to try and encode. In short, at this level, evolution has nothing in common with the species it has placed so overwhelmingly at the top of the heap. My belief is that we can take hints from this tireless mechanism because it shows us that while being empty of intention or understanding it can produce enormous complexity and coherence. Given the right kind of programming I cannot see why a computer, just as tireless but able to operate at vast speeds, should be unable to generate similar complexity and coherence. The task of trying to understand what makes us human in order to build something similar out of silicon and binary digits is not just futile, it’s unnecessary.’