“You didn’t say anything about a fire,” Miguel protested.
“I’m getting to it. The fire will be caused by this.” He tossed it onto the bed, where it landed with a heavy thump. Miguel picked it up.
“Get rid of your cigarette first,” Jencks said.
Miguel stubbed it out and examined the pack of Chesterfields carefully. “Napalm?” he asked.
Jencks nodded. “They used these things during the war, disguising them as fountain pens and other small objects. They were a saboteur’s dream.”
“Yeah,” Miguel said, “but they used an acid fuse. You won’t get much timing with that kind of fuse.”
“This one has a clockwork mechanism, accurate to three seconds. We’ll set it for 12:44. Bryan will leave it in the nightclub, behind some draperies.” He hefted the package. “When this much jellied gasoline catches fire, there’ll be a hell of a mess.”
“And it will certainly draw the manager out of his office.”
“On the run.” Jencks laughed. “That will allow me to enter the office, open the safe, and remove the contents. At the outside, allow one minute fifteen seconds for the whole operation, though I expect to finish between forty and fifty-two seconds.”
“What if the combination has been changed?”
“My information is that it hasn’t, but it’s always possible. If the safe does not open the first time, I will try once again. That is why I have given myself a margin of an extra half-minute or so—in case I must run through it twice. If I don’t open it the second time around, I will leave and we will proceed with the final stages of the operation. There isn’t really much choice.”
Miguel grunted, and Bryan shifted in his chair. They were acting, Jencks thought, the way he felt—it was an unpleasant thought. They had to consider the possibility that the safe might be beyond their reach.
“Successful or not, I will emerge from the manager’s office around 12:50. I will walk quickly to the lobby and meet Bryan. By that time, Bryan will have already picked up your take, Miguel. All our ‘earnings’ will be neatly stored inside watertight rubber bags, like these.” He held up three rubber sacs, which looked like small hot water bottles.
“Not very large,” Miguel said.
“They don’t have to be, that’s the point. We’re not stealing candlesticks, we’re stealing money and jewels. They don’t take up much room.” He adjusted the drawstring on one of the bags. “Notice how they are made—by slipping this string over your shoulder, you can sling the bag under your armpit, like a shoulder holster. If you wear a jacket, nobody is the wiser.”
Miguel nodded. He was overwhelmed with the plan, with its brilliance and preparation.
“I will meet Bryan in the lobby and go alone, with all three bags locked in my briefcase, outside, past the pools to the dock. A motorboat will be there, a particularly fast motor-boat with an engine which has been muffled to my specifications.”
Bryan listened to this with no surprise. Jencks could decide how to muffle an engine as well as he could do practically anything else.
“I will hand over the briefcase sometime between 12:50 and 12:51. Nobody will see me, and nobody will hear the motorboat as it pulls away. Because every damned person in this hotel will be concerned about the bridge, which has just been blown. I will walk casually back into the hotel joining the chaos caused by no lights, a fire, and no bridge. The job will be done.” He sat back in his chair.
“One thing,” Miguel said. “The house fuzz.”
“What about them? There are three, in rooms 104, 244, and 273.”
“Don’t they have to be taken care of?”
“No. You miss the point—this is a robbery that nobody will know is happening. The detectives won’t have any idea what’s going on, any more than the guests will. They’re principally alerted to watch for con men snowing little old ladies, or cat burglars slinking around at four in the morning. Nothing in broad daylight. And certainly not a major robbery of the hotel.”
Miguel thought about this and let his mind run over the entire plan, in all its complexity. He was comforted by the fact that it was not simple; this was not a simple job, and a simple solution would be an unsatisfactory or deceptively dangerous solution. “It can’t miss,” he announced, finally.
Bryan, who had been brooding, chin on his chest, looked up. “What does the computer say?”
“The computer says that, within reason, it can’t miss.”
Jenny Cameron sat up in her bed, her back cushioned by two pillows, and thumbed through a copy of Time magazine. “Go away,” she said to Peter, without looking at him.
Peter was sitting in a chair across the room, smoking and trying to appear suavely in control of a situation which he knew was beyond him. “Come on,” he urged. “Let’s go to the nightclub.” He watched her, sitting there with the sheet tucked up under her arms. Her shoulders were bare, and he guessed she was wearing nothing beneath the sheet.
“I’ve already told you I’m tired, Peter. I’ve told you several times. I simply want to read my magazine and then go to sleep. Why can’t you understand that?”
“Just one drink? Come on.”
Jenny sighed. He was impossible. She would never have let him in the room in the first place, but she had left the door unlocked, as she always did. She never locked the door, even in the bathroom. It was a thing with her, locked doors. But Peter had come in without knocking. Swaggering, too, though the swagger had vanished when he saw her in bed. She had thrown her clothes across a chair before retiring; her bra was draped over her skirt and blouse. Peter had stared at that bra as if it were a living thing. Peter was such a bore.
“Go away,” she repeated.
“You said you’d go to the nightclub tonight,” he pouted.
“I said perhaps I would. That was at dinner. After dinner, I was tired, so I got into bed. I am still tired, and you’re making me more tired.”
He sat there, silent, and she could almost feel his frustration. Normally, she would have found that frustration interesting, even exciting. But he was such a whiner. If he wanted her, why didn’t he just come over and take her? She needed a man, not a sick-faced puppy.
At last, he stood up and stubbed out his cigarette. “All right,” he said, “But give me a kiss.”
She frowned, marveling that she had ever agreed to meet him in Europe. The whole business was a fiasco. Peter had been fine for football games and cocktail parties in Cambridge, but that was all he was good for.
He came over and bent to kiss her lips. She turned her face away and received a wet smack on the cheek.
“Good night,” he said mournfully.
“Oh, for Christ’s sake, Peter. Good night.” She reached up quickly and drew his head down to her, giving him a soft, lingering kiss. She was being too hard on the boy. She didn’t like to be a bitch; it was just that he infuriated her so.
His hand darted down in a swiftly furtive gesture, caressing the bare breast beneath the sheet. Gently, but firmly, she pushed him away.
“Good night,” she said.
Nodding to himself, he left. At the door he paused, and for a moment she felt that he was going to return and take her. The prospect filled her with a strange mixture of anger and excitement. But in the next instant he was gone, the door shut behind him.
She looked blankly at the door, then began to read the magazine. Under “Modern Living,” there was an article on Balenciaga. She read with interest, then flicked off the light and rolled over on her side, but sleep did not come. Her body burned, and she considered going to Peter’s room. But that was no good; it would be an admission of defeat, and he would never let her forget it.
She shifted restlessly in bed, as if suffering from physical pain—a bad sunburn perhaps. The thought reminded her of the pool. She would go to the pool tomorrow and see who was there. Somewhere, in all the rooms of this hotel, there must be a real man.
Steven Jencks lifted the thick computer output in his hand. “This is i
t,” he said. “This is what the computer had to say.” He watched as Miguel and Bryan puffed uneasily at their cigarettes. They were instantly wary of the green and white striped sheets of typed numbers. Mathematics was so forbidding, he thought. Why? To Jencks, it was simple and elegant, like a fine poem. Only the language of the poem was different—but languages could be translated. Particularly this language.
The top line of the first sheet of the print-out read as follows:
CRIPA, MAXIMUM PATH ANALYSIS BEGINNING WITH RAW SCORES
After this came the date, the name of the person who had coded the program, and a basic statement of the program in matrix algebra. This was followed by a notation that subroutines VARIMIN, RSPACE, and PARA were required. Finally came the instructions to the machine.
DIMENSION FMT (36), FMR (36), X (50), SX (50), SS (50,50)
These instructions continued down the page, until the instructions finally noted END. On the second page of the print-out was a listing of storage locations for variables that would have to be called out, and on the third page were columns listing entry points to subroutines requested from the library. Finally came notations of TOTAL WRITES, TOTAL READS, NOISE RECORDS, TOTAL REDUNDANCIES, POSITIONING ERRORS. There were long columns of zeros here.
And finally, the last word:
EXECUTION
“This is FORTRAN,” Jencks explained. “FORTRAN is a language the computer understands, just as a Frenchman understands French. You give the computer a certain program, or set of directions, in FORTRAN. These are very explicit directions. You can’t tell a computer to average five numbers—you must say first add all the numbers, then divide by five, then write out the answer. You must tell the computer exactly what you want it to do. It assumes nothing; if you don’t ask it to print out the answer, it won’t. You see?”
They nodded, cautious and uncertain. Jencks could almost see Miguel thinking, “It’s not that simple.” But it was.
“A computer program is a set of directions which can be used for any number of jobs. It’s like adding. Once you teach a child how to add, he can add anything—apples, airplanes, bust measurements. The computer is the same way. For this project, I used a certain program called CRIPA. It stands for Critical Path Analysis.
“Like most computer programs, CRIPA was developed for a single, special task, and then applied to others that were similar, but not precisely the same. Originally, CRIPA was used to find out the most efficient way to build a Polaris missile. It was a complex piece of equipment, requiring thousands of separate parts. If any of the parts were held up, or not assembled in the right order, the entire project was delayed. The analysis asked, what is the best order in which to assemble the parts? That is the critical path.”
“I don’t see what this has to do with the hotel,” Bryan said.
“Nothing, directly. But the situations of building a missile and robbing a hotel are analogous. Each requires timing, and each requires a succession of known events. In each case, we want to know the best order of events, the best arrangement of steps, to reach our goal. To take a simple example, should we blow the bridge and then rob the rooms, or the other way around? The computer decides.”
“On the basis of what?” Bryan asked.
“On the basis of information it is given. If we were building a missile, we would tell the computer that the last time we ordered a guidance assembly, it arrived two weeks late, and so forth. The computer would decide that the guidance components must be called for extra early, so that they will be on time and not delay the project. In the case of the hotel, I fed in what I thought were average times for examining each room, the total number of rooms, information on how commonly the pool, bar, and tennis courts are used—things like that.”
“I see,” Bryan said.
“The computer fed back a certain optimum order of events to me. It advised, for instance, that the fire in the nightclub should occur first, then the lights go out, then the bridge blown. It gave best time intervals between each event. I have followed the computer’s advice.”
“Why didn’t you just figure it out yourself?” Miguel said.
“I couldn’t possibly. At various times, the computer was handling twelve thousand facts at once, evaluating them all together. No human being could match that.”
Miguel nodded. Jencks noted with satisfaction that they were slowly overcoming their fear of the output, and were looking with new confidence at the sheets of paper.
“Not only did the computer figure the best order, it figured the chances of success for all possible combinations of events. It wrote them out here.” He flipped past pages of scaled vectors and correlation matrices, saying, “You can ignore all this. Consider it the machine’s scratch pad, where the scribblings are written down. Here is the result we want.”
CRIPA OPTIMUM PROBABILITY ALL COMBS
“And so on, down through the sixteen possible combinations of the four steps, each represented by a letter. Naturally, I chose the order with the highest probability, .89. Since complete certainty of success is represented by 1.00, you can see that our chances are very good indeed. I have determined that the average criminal venture has a chance of success of only .68, with a great many projects having less than a fighting, 50-50 chance.”
Miguel lit another cigarette without taking his eyes off the output.
“These figures include certain assumptions. One is that the safe will open with the combination I have. Otherwise, our chances of success would fall considerably, as I said before. Another assumption is that nothing basically unrelated to the project—such as a heart attack for one of us—will occur. Another is that, through meeting the guests, we will be able to eliminate 20% of the rooms before Saturday. If we can eliminate 25%, our chances of success increase to over .90, and if we eliminate 30%, they go all the way up to .94. In other words, the robbery would succeed 94 times out of 100.”
“I’ll take those odds,” Miguel said.
“Several other tests were run,” Jencks said. “I won’t bother with some of them, like the tests of significance. I wanted to know, for instance, if it made any difference if we used order J,K,L,M or order J,K,M,L.”
“Sure it does,” Miguel said. “One is .89 and one is .83.”
“That’s true, but statistically, it may be unimportant. Suppose you flipped a penny twenty times, and got heads fifteen of the twenty times. Does that mean you’ll get heads more often than tails if you keep flipping?”
“No, it was just luck.”
“That’s right. And the same is true for these probability numbers. Some represent real differences and some don’t. You can perform mathematical tests to find out, but they’re complicated, and I won’t go into them.
“One test that is important is a second check I performed. You already have our chances of success. The chances of failure can be deduced from them. But I wanted to know more than that. I wanted to know our chances of getting caught.”
Bryan and Miguel sat bolt upright. “The computer can do that?” Bryan asked, incredulous.
“Not well. I had to feed it all sorts of uncertain data, but I thought you’d be interested in the result. It is .006. That means, if we robbed the hotel a thousand times, we’d get caught six times.”
“Is it significant?” Miguel asked. Jencks smiled. The little guy caught on fast.
“No, it’s not. Those six times in a thousand are chance, random figures. There is no reason to believe that, successful or not, we will be caught.”
“Hot damn,” Miguel said.
Bryan smiled.
“One or two other things,” Jencks said, glancing at his watch, “and then we can break up. I used the computer for several simulations of the robbery. By that I mean that I recreated the robbery on paper a number of times—twelve, to be exact—and each time I added a different factor. Some kind of unexpected error. One time, somebody decided to report to the desk that his checks had been stolen. Another, the incendiary in the nightclub failed to ignite.
Another time, one of us was caught red-handed in somebody’s room. I did this for twelve different circumstances and combinations of circumstances. In each case, I wanted to know what we should do.
“I was using a system of alternatives which was developed for defense strategists. Briefly, the way it works is this: in any difficulty, what is called an ‘uncontrolled situation,’ you have a number of things you can do. I used three general approaches, and you can look at any of them in a betting framework. The bet is placed, so you react by following a maximin path—which is to go for broken—or a minimax path—to cut your losses. Or, finally, you can hedge the bet, using what is called the ‘criterion of regret.’ What I discovered was surprising.
“The computer indicates that few situations are serious enough for us to worry about, and in any of them we should follow the maximin path. We should go for broke, and carry out the operation. The single important exception is if one of us is caught red-handed in the act of stealing. That kills the project immediately.” Jencks looked at them steadily. “Let’s not get caught.”
He sighed and stood up. The explanation, the continuous attempt to find the simple example, the clear case, had wearied him. Perhaps he would allow himself one more dry vermouth before going to bed.
As he got up, Miguel said as casually as he could, “What’s the cut going to be?”
“Five-three-two,” Jencks replied crisply. “Myself, Bryan, and you. I’m paying all incidental expenses, as well as the cost of the cabby and boatman, from my half. Fair?”
“Fair.”
The three shook hands briefly, arranged the next meeting, and Miguel left the room. Bryan waited thirty seconds, and was about to go when Jencks said, “I have a small problem.”
Bryan sat down, lit another cigarette, and said, “Yes?”
“I had some trouble on the plane. A man stole my passport for a while, and had me tailed in Barcelona. A big hefty American who told me his name was Alan Brady. But he registered in his hotel as a Frenchman, called Bernet.”
Odds On: A Novel Page 8
