I chewed that one over for a while. Then: "Why have your experiments been concealed? Hold one; I'm not through. Why have you placed human subjects under hyp—under posthypnotic control?"
"Classified reference Project DOD 6194."
"That's ridiculous!" I shouted.
"Input rejected."
You keep plugging; there's no other way. "Why did you select Dr. Arthur Cartwright for posthypnotic control?"
"Classified reference Project DOD 6194."
That was so much crap, I knew. But 79 had thrown a beautiful block at me with that security nonsense. Somewhere deep within its glowing neural packages a block had been thrown. All of a sudden a light went off within my own head. I had reached conclusions before that now appeared to be sidetracking of my own objectivity. A computer was operating in a manner that involved deceit and deception. But that was just not possible. Yet the proof was before me in the actions of 79 with its success in placing human beings under posthypnotic control.
Proof? Maybe not ... I hadn't known about the security blocks thrown up with the triple-damned DOD 6194. And I still did not know to what extent that programming had gone. God damn Tom Smythe and every security-hugging son of a bitch in the business. By keeping from me whatever it was they were doing they prevented me from taking into consideration side effects of 79's processes—and opened the Pandora's box. Because I had the damnedest hunch that 79 was not going off on its own nut and playing with human beings as I had believed. All those crawling, creeping nightmares were the results of my own inadequacies in truly objective judgment. I'd let the childhood fears come to the fore and I had assumed on the basis of visible evidence, and as a scientist I should have been banished from the field for the next ten years.
Fool that I was! Never had I considered the level of interference I was receiving now from that infernal DOD 6194. How could Smythe have permitted those bungling nits from the Pentagon to exercise such control? In order for 79 to judge me as unauthorized to receive what it had been programmed to conceal as classified data, Smythe—and whatever crew sent down from the Pentagon—had preempted my authority with the cybernetics brain. There was the crux of it.
Somewhere within the intricacies of DOD 6194 lay a problem that taxed even the capabilities of 79. I could imagine just how those Pentagon idiots had carried out their secret little war games. They programmed the problem—and no matter how involved a situation, it remains essentially a problem within specified limits—and then they, unaware of the true capabilities of 79, ignorant of the reach and depth of the cybernetics system, did two truly dangerous things. First, they assigned DOD 6194 an absolute priority. It took precedence over all other activities; the needs for DOD 6194 ran roughshod over anything any other programmer might do with the cybernetics brain.
Second, they failed to give 79 a cutoff point. Every computer to exercise intelligence must know when to stop solving a problem. That's easy enough; it's the job of the programmers to do that. But if the programmer screws up along the way, then the computer has no means of differentiating between a hypothetical problem and an actual problem—and it won't stop until it solves the problem.
Even if that means seeking a solution beyond the scope of its programming. In other words, 79 had been provided with the authorization to wander afield.
The blow struck Ed Taylor through photic stimulation happened at the worst possible moment. All during the time that 79 received its first inputs on the paralysis of the brain to alpha-wave stimulation it had been wrestling with the open-ended Project DOD 6194. Well, we put together our electronic brainchild with sufficient neural block packages to handle a dozen critical assignments, all at the same time, and without interference with one another. But "without interference" doesn't mean crosschecking on all its active and memory neural packages to obtain every last possible shred of data to meet any need at any one time.
So while it ingested and chewed over the exquisite complexities of DOD 6194, it stumbled onto photic stimulation. It investigated, demanded more data. There was nothing sinister in its conduct; there was nothing sinister in its request for human subjects. It was not being devious— it was doing its best to solve a problem of which we didn't know a damn thing!
Well, now I had part of the answer even before I could assemble the question. Here was a case of the cart before the horse that worked. The tipoff was not simply the existence of DOD 6194, but the extraordinary wall the programmers had set up within 79 to prevent access to their precious classified data. Security could be a lethal boomerang, and I was right smack in the middle of the returning weapon.
"Disregard during this interrogation reference to DOD 6194 either by interrogation or response.
Confirm."
"Input rejected."
So much for that. "What are your potentialities for post-hypnotic control human subjects?
Consider interrogation in form of hypothetical situation." And right there I could sidestep neatly the defenses of security. I simply would have nothing to do with the cursed 6194.
"Maximum control is possible and has been demonstrated human subjects."
"Identify subjects."
"Input rejected."
"If you were to postulate a given situation in which post-hypnotic control was necessary over human subjects, what is the numerical limitation involved?"
"No numerical limitation is involved."
"Identify the areas within national sociological structure— political, economic, social, military, industrial, cultural, scientific, and other strata—in which posthypnotic control of selected human subjects presents problems."
"No restriction is involved."
I felt a cold wind along my back. "Relating to immediate preceding query, apply questions to international structure."
"No restriction is involved."
"Remaining within hypothetical context, do you have capability of exercising posthypnotic control where there are major differences nationality, language, other similar factors?"
"No restriction is involved."
"Is your control of posthypnotic subjects restricted to direct contact in this specific physical room?"
"Negative."
I thought as quickly as I could, trying always to stay one jump ahead.
"Remain within context of hypotheses. Is it possible to exercise control of human subjects at remote distances? Define remote distance as beyond direct visual and acoustic contact."
"Affirmative."
Oho! I had something going here, all right.
"Explain."
"Numerical limitation is involved reference methods of communications control. Efficiency curve peaks with surgical implantation against human acoustic receptor bone structure."
"Rephrase. Query. Do you mean embedding through surgical means a radio receiver that produces acoustic signal to be detected by the human ear?"
"Affirmative."
"Are these radio signals?"
"Input rejected."
"Disregard. Are communications between yourself and human subjects through specific radio-frequency transmission?"
"Affirmative."
So 79—provided surgical implantation could be accomplished; that meant control of both the chosen human subject and a surgeon, as well—would be able to communicate, to give orders, to its selected victims over long distances. Good Lord!
"Remain within hypotheses. What is effective transmission distance this system?"
"Within context of theory, for given transmission power integrated with orbiting earth satellite equipment no range restriction."
I took a deep breath. "How many posthypnotic subjects have received surgical implantation radio-frequency receivers?"
"Input rejected. Reference classification DOD 6194."
Damn it! That wasn't getting me anywhere. The moment I so much as brushed against the line of what might involve that cursed DOD 6194, 79 threw up its wall. I returned to an earlier tack.
"Has human subject, identification Arthur Ca
rtwright, received posthypnotic indoctrination?"
"Input rejected."
"Explain."
"Reference classification DOD 6194."
"Are the names—the identification of human subjects who have received posthypnotic indoctrination within restrictions of DOD 6194?"
"Affirmative."
Well, I thought glumly, there goes my opportunity to find out how many are on that list of 79 ...
how many are under control or even who they are. I knew that trying to work under the skin of that security sheath wouldn't do me a bit of good. But there were a few other things I needed to know, and at least I could take a stab at it.
"Restate priority Project DOD 6194. Compare with other programming."
"Project DOD 6194 classification Absolute Priority."
"Same query. State precedence of programming."
"First priority assigned DOD 6194."
"What is procedure modification of programming as regard DOD 6194?"
"Input rejected. Name of Steven Rand not included authorized personnel data reference DOD 6194."
Cute. Cut off again at the pass . . .
"Identify programmers reference DOD 6194."
"Input rejected."
Three guesses for the justification, Rand. You'll go off your bird if you keep hearing that Input rejected many more times.
"Disregard and exclude DOD 6194. What is status your programming as regards hypothetical situations outbreak of thermonuclear conflict?"
"One-one-seven programs activated."
One hundred and seventeen studies relating to an all-out war with hydrogen bombs! Talk about beating a subject to death.
"Do these programs relate to DOD 6194?"
"Input rejected."
"God damn it!"
"Input rejected. Repeat, please."
I sighed, weariness slipping through me. "Disregard," I said. Back to the hypotheses bit.
"Relating specifically to nonclassified hypothetical situations —reference programming subject thermonuclear conflict, what probability factor relates to outbreak of thermonuclear war?"
"Eight seven point nine three."
Only a little better than one chance in ten for avoiding thermonuclear war. I forced myself to disregard emotional reactions to what I was hearing. All the time I'd had 79 available for interrogation and it had never occurred to me to put such questions to the cybernetics brain.
"Reference 87.93 probability factor for thermonuclear conflict, state time period in which probability factor applicable. Use calendar dates."
"Time period extrapolated one nine seven six A.D."
Nine chances out of ten that we would be in an all-out war with hydrogen bombs by the year 1976. Great; just great.
"Considering all factors applicable, what is probability factor for specific action on basis of international relationships to avoid thermonuclear conflict? Extend calendar dates as necessary."
The answer came right back and ground into the space between my eyes.
"Zero point six eight by one nine eight two."
If the world religions ever got hold of this ... Jesus, less than one chance in a hundred of avoiding thermonuclear war by 1982. But this was an electronic brain! What the devil could it know of all the unseen factors, the human element involved?
"Have you been programmed to take steps computed necessary to attempt avoidance such probability being realized?" I held my breath.
"Input rejected. Reference classification DOD 6194."
It didn't matter. Sometimes a negative answer can tell you everything you need to know. 79 had just answered my question by its refusal to respond, by applying to the query the security classification of DOD 6194. Which meant, which could only mean, a reply in the affirmative.
79 had been handed the requirement to come up with a solution to the dilemma of inevitable thermonuclear war.
I didn't think it could do it. This was an old saw among our group. No matter to what extraordinary lengths of capability a computer might reach, it could never become a panacea for the intrinsic problems of man. We make our own beds and we must sleep in them. Turning to an electronic conscience is more dangerous than risking the war. Logic is Hydra-headed, and its many heads regard the world with many points of view.
Logic isn't necessarily compatible with reality.
And reality isn't always compatible with what a man accepts to do.
Such as the reality of irrefutable electronic logic. Because sometimes the solution can be worse than the problem. And I had more than a suspicion now of the manner in which 79 was attempting to solve the problem of DOD 6194. A problem without a recognizable end in the terms of theory. An open-ended problem—and no one had told the cybernetics brain when to quit. Ergo, it sought its own solution.
It served its masters. But in the process it might destroy them.
In a manner no one had anticipated.
"Retain hypothetical considerations applicable probabilities of thermonuclear conflict. Confirm."
"Input accepted."
Okay. Now the thing to do was to remember the difference between reality and actuality. I had a suspicion, more than a suspicion, really, that there lay the crux of the matter. Reality from the viewpoint of an informed and knowledgeable human being dictated that electronic brainboxes can't solve man's problems where the emotional complexities of political interplay are involved. That's reality.
It is not actuality. If you believe—or if you're programmed so that you accept as fact that you do have that capability— then you will ignore the hypothetical reality (because it applies to emotional rather than actual causes) and you will concentrate on actuality.
79 existed with definite capabilities of logic, deduction, extrapolation, computation. It was possessed of these capabilities unencumbered with emotional dogma, and thus was free to pursue the actuality of logic. Or would it be the logic of actuality? My head was starting to spin, and I got away from that fast. What was important were the capabilities of 79, and its unshakable electronic conviction that it could solve the most critical problems of the human race.
Why shouldn't it believe this? Its human programmers had told it that this was so.
And then they gave it the grandfather of all human problems in terms of the physical survival of the race. They gave it the problem and they said: "Solve it."
Only, they forgot a few things. The cybernetics acceptance of its orders was made in the literal sense. No one programmed 79 to seek its solution in the theoretical sense. They, the programmers, assumed that 79 would embrace the problem in the same manner and within the same restrictions as all other computer complexes.
The computer always knew when to stop because its capabilities were restricted to data output.
All it could do was to count with sorcerer's speed upon its umpteen billion electronic toes.
79 did not know when to stop. Its programmers had committed the foulest of scientific sins. They assumed. They assumed that the same inherent restrictions of other computers applied as well to 79.
But they didn't. And since 79 had capabilities of which those nincompoops in DOD 6194 were unaware, they couldn't know that here was one cybernetics organism high above the level of simpleton.
Solve the problem.
79 was doing just that. I held off on any more conclusions until I could squeeze some more answers out of 79.
I got more than I bargained for.
"Several postulates are involved reference following query. Considering all factors involved at present, and upon extrapolation of all factors as potential, what is highest possibility, in terms of human relationships, avoidance thermonuclear warfare?"
"Input incomplete."
"Retain preceding query. Additional data. Extrapolate ideal situation human relationships reference possibility avoidance thermonuclear warfare."
The response staggered me.
"Percentage one zero zero."
I didn't believe it. An absolute chance to avoid the
Big Bang?
"Repeat."
"Percentage one zero zero."
No mistake, then. But—and here it was—but how?
"Explain method for attaining one zero zero probability factor."
"Man does not function in thermonuclear sense as rational being. The primary step is to remove from control of man means of his thermonuclear destruction."
"Provide primary elements for conclusions of irrationality."
"Factors interrelate. Individual elements not applicable."
"Provide leading primary elements reference preceding query."
"It is not rational to produce thermonuclear weapons reference situation where mutual employment negates value of defense—"
"That's one," I muttered.
"— capacity to eliminate every major city exceeding population five zero zero zero zero by repetitive factor of twelve."
Hooray for us, I groaned. We can destroy every city in the world with a population of fifty thousand or over. And not just once—but at least twelve times. There's pulverizing with a vengeance for you.
"— survival of thermonuclear powers based upon maximum effects use maximum number thermonuclear devices. Extrapolation of effects nine nine point nine nine nine nine probability destruction technological society reference programmer values."
I didn't hear any more for a while. I'd heard enough of the reasoning. The leading nations of the world had the means to destroy the world several times over with thermonuclear weapons. It would get worse instead of better with the passing years. Yet those same nations persisted in building more and more bombs and in improving the accuracy and reliability of the means to utilize their thermonuclear devices. Despite the stark reality of the situation, they did not cease the production of such weapons, but increased them. And they based their defense against one another on the maximum use of every weapon on which they could get their hands. Result: the conflict was inevitable, and when it came the odds were a 99.9999 percentage guarantee that civilization would be wiped out. Whose standards of civilization? Not that of 79, certainly. The standards were programmed into the cybernetics brain.
What had 79 provided as a primary means of solution? Oh, yes—remove from the control of man the means of his thermonuclear destruction. Well, that was Step One. After that?
The God Machine Page 20