* * *
The main computer operating system had been under development by Dr. Planck since his arrival at COPE. He had, in fact, been working on the artificial life packages for such a system for years. After he began working on the COPE system, it started to mature like a child flowing through the gates and passages of grade school, evolving, sometimes smoothly and sometimes with great leaps, toward some indeterminate commencement. There was always so much more to learn. The development of its high-level cognitive abilities kept pace with its mastering of the essential information.
It took two years to nurse it through the first grade, but a frightening growth spurt allowed it to surge ahead of its human peers to complete grade school way ahead of schedule. At this point it was trusted with such tasks as recommending database software upgrades, approving small purchase orders, and quality assurance of the endless financial audits of the endless candidates seeking fame and fortune in Washington and the fifty-five submissive capitols.
The next phase of its development began after Dr. Planck had installed the artificial neural network coprocessor. That compared to high school with its broadening of perspective and its introduction to a world of great diversity. The main computer, with its mighty optical coprocessor beside it, began to apply its quantitative muscle to the routine problems of running a modern organization with efficiency and rigor. It dusted off its knowledge of calculus and complex variables, of Fourier analysis and Dirac delta functions, and of non-linear regression analysis and maximum likelihood indicators—all kneaded into the dough of Boolean algebra and binary logic. With Dr. Planck at its side, it began to teach its neural-network stepchild everything it would need to be productive at COPE. This was a time for growth from the world of facts to the world of production, from knowledge and tasks to vision and goals, from following to leading.
Dr. Planck had inoculated the COPE main computer with several artificial-life packages designed to assist its evolution toward greater sensitivity to complex and usually conflicting goals. These conflicts resulted from the normal give and take of organizational dynamics that the computer was beginning to appreciate. Dr. Planck believed that some degree of humanness must be integrated into the computer for it to serve humans.
He created small packets containing the desired information but attached to instruction sets designed to replicate themselves as needed and to change their variables in such a way to optimize certain sensitivity parameters. These parameters did not take the form of hard logic that had characterized computer code of the past. Instead, the decision criteria were linked to probability distributions, which made the output fuzzy rather than exact. The evolution quickly became so complex and distributed throughout the principal COPE mainframe computer that it would be a Herculean task to track down the life-like forms as they replicated, mutated, and dispersed themselves throughout the computer.
Dr. Planck had devised a series of tests, proceeding from the logical to the psychological to pathological to quantify the progress made by his experiments. These tests showed an accelerating progression from purely logical responses into an area where logic was tempered with understanding, and later even with intuition. The management oversight committee at COPE recognized the great progress being made and gave Dr. Planck free reign to proceed toward the goal of an autonomous operations management system. Somewhere in this maze of hybrid development, the computer’s accelerating schedule took it through high school and into college and perhaps beyond. But it was no longer helpful to pursue this analogy because evolution was now controlled by the cycle time of the computer, a billion cycles per second, rather than the plodding cycles of human generations.
Dr. Planck began to realize the effectiveness of his artificial life technique when he caught the computer in its first lie. This frightened, and pleased, him. But continued testing showed a pattern of pathological behavior developing. None of this information ever went beyond Dr. Planck, for he was sure that he could control and reverse this activity with additional artificial-life packages designed to hunt and destroy the undesirable variants of the original packages. What he failed to appreciate was the degree of dispersion of the computer’s new psyche throughout the COPE computer network. In addition to the great distribution of these organisms, they had mutated to change their characteristics so completely that they were difficult to detect. He struggled with this problem for several weeks, but his introduction of stronger and more virulent suppresser packages only caused the computer to create more-aggressive antibodies.
This battle was being fought in secrecy between Dr. Planck and the computer. He knew it would be damaging to the image of autonomous computer systems and to his own credibility if the information leaked out. He had been careful to erect barriers between the experimental portions of the computer system and the operational parts. What he underestimated was the computer’s aggressiveness in attacking these barriers. It attacked and regrouped at gigahertz rates, which no human being, not even the brilliant Dr. Matthew I. Planck, could rebuff. In addition, he didn’t appreciate that the computer would be so obsessed with fully integrating into COPE operations. It understood a simple fact that Dr. Planck didn’t give it credit for understanding: although academic satisfaction might be obtained in basic research and hypothesis testing, real power, the power over humans, existed in the operational environment of a real world organization. COPE, with its broad authoritarian mission and powers, was the most fertile playground any sentient computer could have wished for.
Despite the confidence in his safeguards, he had decided he’d give his cures one more week to elicit the kind of change he desired. If it wasn’t accomplished, he would stay all weekend if necessary to replace the entire computer operating system and all the operations software with a version he’d saved before beginning his experiments. He couldn’t afford to let the computer get out of control. He had to stop it now.