The Idea Factory

by Pepper White

  "Of course, Koosta's an old man now," Doc said.

  "But has he lost his enthusiasm?" I asked.

  "Nope.'

  "That's the important thing, isn't it?" I added.

  "Sure is. You read French?" Doc asked.

  "Yes, I do," I answered and started reading the text of the letter.

  "It looks like they want me to fill out this questionnaire for some book they're doing about their ship's voyages," Doc said. "What's this question?"

  "Um, I think they want to know how many times you went on the ship's voyages. If it's one or less they don't want you to fill out the form," I translated.

  "Think eight'll qualify?" Doc quipped. "How about this question here?"

  "They want copies of any photographs you may have taken while on board."

  "We've got two thousand. Wouldn't they have a fun time picking through those to find the best ones?" Doc said.

  While we were on the subject of the sea I asked him when he would raise the Titanic.

  "We know where it is," he said as he pulled out a crumpled piece of paper from his wallet. "It's at 41 degrees 16 minutes north, 50 degrees 14 minutes west. I got that out of a book I was reading. That's an area 50 miles by 50 miles. That's just a point on the globe, but try finding a ship in that much water. It's not easy."

  Doc continued. "We found her sister ship on the bottom of the Mediterranean some years ago, but that was in only 300 feet of water. The Titanic's in 12,000 feet of water and my sonar only works to depths of 3,000. He turned around and drew on the blackboard beside his desk.

  "So you've got a lot of cable, and it's always going to be hanging behind the ship, making the hypotenuse of a right triangle. That's no easy trick, to have 15,000 feet of cable dragging behind you in the open ocean, then to coil it and uncoil it. And if you find it, then whadaya do? It's a whole 'nother can of worms to try and raise it from the bottom."

  This Herculean, mythic task was reduced to geometry and seemed doable the way Doc talked about it.

  Bill had found the circuit and brought it in to Doc. "I only could find one, Doc," he said.

  "Well, give it here; I'll make him a drawing. He's a grown man; he can build the circuit himself," Doc snapped.

  Doc sketched the circuit in a minute. I ran down the hall to make some copies, returned, and gave him the original.

  Bill tied the cords from the power supply box with some small wires attached to the box handle. "It's a patented method I've developed to prevent people from tripping over the cords," he joked.

  As Doc and Bill loaded my arms with the equipment, Doc issued the conditions of the loan. "We don't like to lend to people who aren't successful, so work hard now and take some good pictures," he mandated.

  "I'll have some for you in a week, sir," I answered.

  I turned the corner to Building 13 and Doc exhorted again from his office, "To the mark!"

  May 25

  I set up the Polaroid camera and Doc and Bill's flash. Chet walked in and said, "What are you doing? We're only taking movies here. What are you wasting your time for?"

  "Look, Chet, I just want to give this a try. I wanted to see about making faster movies and Professor Edgerton suggested I try this instead. Who knows, we might learn something," I answered.

  I felt a little more confident, a little more able to talk back to Chet since the time I was right about not mounting the fuel injector.

  "Well, all right, but don't spend more than a day on it. You gotta make movies of combustion and take the pressure data and analyze it. We gotta get you outa here and into the real world," he said.

  Look, Chet, I know you don't think I'm Ph.D. caliber, but let me enjoy following my intuition once in a while for the duration.

  "You need any help?" he asked.

  "The key is the timing. I need the fraction of the millionth of a second that the flash is on to occur within the thousandth of the second that the injection lasts," I said.

  "Just use the timing boxes we have for the rest of the experiment," Chet said. "Put the strobe flash on one timing box and put the injector on the other timing box, and dial in the delay just as you did with the movie camera."

  The high technology of mechanical engineering is in precise space. The high technology of electrical engineering is in precise time. The question remained, How many thousandths of a second delay should we put between when the injector starts and when the flash starts? The injector took about 42/1,000 of a second to inject from the time the timing box told it to start, so I suggested that number to Chet.

  "No," he said, kind of arbitrarily. "Let's start with 40."

  We dialed in 40. No fuel spray on the Polaroid film.

  "Now try 45," Chet said.

  Still no fuel spray.

  "Forty-three." No fuel. "Forty-one." No fuel. "Forty-four." No fuel.

  "Forty-two." A crisp sharp image of the fuel displayed itself on the Polaroid print. Each of the five jets looked like a tiny tuft of fluffy white cotton. It was beautiful, clearer than the movie. Doc was right.

  C H A P T E R

  17

  The Joy of Six

  11. . . of the tree of the knowledge of good and evil, thou shalt not eat of it. . . ...

  Genesis 2:17

  Schedule:

  Summer '83: 6.001 Structure and Interpretation of Computer

  Languages (Siebert)

  6.931 Basic Electronics (Zapf)

  2.996 Thesis

  May 27

  Ari and I went to graduation to hear the speech by Helmut Schmidt. It was a cloudy day for the professors in the procession with their academic robes and their hoods from Stanford and Oxford and Cambridge and MIT, and their scepters and their funny hats. It all seemed so medieval, as if the wizards were displaying their power.

  Schmidt marched past me, with four younger men in plain black caps and gowns marching on all four sides of him. The four younger ones didn't look like scholars; there wasn't the focus and intelligence in their eyes. And instead of looking forward like the people in funnier hats, they looked up at the rooftops, at the second, third, and fourth floor windows around Killian Court. Schmidt scanned the periphery as well.

  The five looked tense. I wondered whether they were mentally reviewing the choreography if there were an attempted hit: Schmidt would hit the deck first, then two would jump on top of him to cover him, and the other two would whip out their Uzis from under their robes and open fire on the would-be assasin. Surely the security sweep had sterilized the area, though, and the bodyguards were only a redundant system.

  They reached the stage without incident. A few of the older parents in the audience softly sang "The Star Spangled Banner," and I sang it loudly the way I had ever since I'd learned the words in Cub Scouts. When the anthem ended, a guy in the row in front of me said, "Play ball."

  Midway through Schmidt's speech, the sounds came from behind the stage. "Click, thunk." The scroll fell from the frieze between the Ionic columns of Building 10. "NOTHING'S."

  "Click, thunk." "IMPOSSIBLE."

  Score one for THA. Nice work, Ghan.

  Course Six is the Electrical Engineering department. This is where the power is. Paul Gray ('54), president, got his start as a professor in Six. Gerry Wilson, dean of engineering, was a professor in Six. And, of course, Doc Edgerton was in Course Six. Course Six invented radar, artificial intelligence, and computers. Course Six is high technology.

  July 20

  Software. Freshman double E's take six double oh one (a.k.a. six double no fun) to learn to program in LISP (LISt Processing, or Lots of Insidious and Stupid Parentheses). This is also where they begin to leave the rest of the world behind them in their ability to solve problems and to make things work.

  I'd heard so much about Six at milk and cookies in Senior House that I wanted to experience it, if only vicariously. Anything that is not graded is vicarious at MIT, because you need the pressure of the other students and the quizzes and the graded problem sets to force you to absorb t
he mountain of data presented. The two week summer version would give me a big enough vicarious taste.

  The underground guide to Course Six noted: "6.001 is not a programming course-it teaches you how to think about complexity." Perhaps it would fulfill Professor Mikic's promise to me in that first week two years ago.

  Professor Siebert was a little younger than Gyftopoulos and Greene; he spoke forcefully, and seemed more like a Fortune 500 executive than a computer science professor. But his manner didn't hinder his ability to transfer the data.

  "Abstraction," he said. "That is the word. When you go about designing a computer program, or designing anything for that matter, you need to make abstractions."

  Right. What's an abstraction? And I thought I was making progress in this place.

  Siebert went on to say that we might already be familiar with the term black box. A thing to which you give an electronic signal or signals is referred to as as a black box. A signal can be as simple as the voltage from a car battery or as complex as a radio wave. The black box operates on these input signals to give you an output signal. It's called a black box because you don't know how the black box performs the operation, but you do know what the operation is. The black box concept was developed in the early days of analog electronics, when they were, in many cases, literally black metal boxes.

  Like those in Doc Edgerton's lab. I didn't have to know precisely how the capacitors and other electronic component devices inside the EG&G strobe boxes worked; I only needed to know what the boxes did and how to interface them with my experiment.

  Abstraction is the word. From abstractus, "To drag away from." Abstraction is the act of considering something as a general characteristic, apart from concrete realities, specific objects, or actual instances. In the language of 6.001, an abstraction is also the general characteristic itself, the function or the operation.

  So abstraction is the process of making abstractions, making black boxes. You divide a heavy, complex intractable problem or system into smaller, tractable pieces by abstraction.

  Once you've dragged the smaller pieces away from the larger whole, and found the general characteristics, or the relationship between output and input for the abstraction, you can link the abstractions to attain a more complex abstraction.

  Yes, but what about entropy?

  "With every abstraction, you need to define an 'abstraction barrier,' i.e., the limits of the abstraction," he continued.

  An automobile can be thought of as an abstraction. It has a weight and emissions and a length and a width that civil engineers and highway planners deal with. But the car itself is a collection of abstractions. You could call the engine an abstraction. The transmission is an abstraction, and the tires are abstractions. Each abstraction has performance characteristics that can be modeled by an engineering group.

  Model. Key word. So an abstraction is like a model. And a model of a system may be composed of linked models of smaller systems, or subsystems.

  "Now in design," Professor Siebert said, "the task will be to define the abstractions and the abstraction barriers."

  You will also need to have clean interfaces between the abstraction barriers; this will enable more than one person to work productively on a large, complex system. It's the old divide-andconquer approach. Any technology consists of linked abstractions, and the idea is to divide the abstractions in a manageable way.

  This is what Chet and Scott and I had done with the RCM. I'd picked up the front half, the fuel system, the clear window design, the nozzle design, and the movie camera, while Chet and Scott had taken the rear half of the machine and all its attendant subsystems. Thus I could invent and tinker in the front half and feel like the old-time inventors, while Scott and Chet did the same in the rear half of the machine. Once our two systems worked, we would choreograph the whole process of the experiment with the timing electronics I had ignorantly assembled.

  This level of subdivision has its limits. It works its way into the real world, where engineers and scientists are divided and conquered by allowing them only to know the cellular information they need to produce their assigned abstraction. As they climb the branches of the corporate tree, they are privy to more and more information, a broader view. They see that the whole thing is pretty easy once they know how to do all its parts, and if the noncompetition agreement they sign is not tight enough they form a spinoff company with some of the other abstractions. As John Delorean did.

  It goes further. Turning a bolt mindlessly at a car assembly plant is an abstraction, easy enough to teach a robot. And a robot will never snort cocaine.

  More Siebert lecture. "We encourage you to think wishfully."

  When you're solving a problem, wishful thinking will help you define the appropriate subproblems. Wishful thinking is simply saying, "Wouldn't it be nice if I had a thing that did such and such a thing." Once you know what you want, it's easy enough to get it. Once you name a spirit, you have power over it. You just keep saying that question over and over again, looking for the answer, breaking the problem down into smaller and smaller pieces. Eventually you get to the point where you can actually solve one or two of the pieces, or one of your subordinates can. The trick with large systems is to choreograph or orchestrate it so that all the problems are solved at more or less the right time and people aren't sitting around waiting for some other abstraction to give them the information they need to do the next thing they need to do. This is what we mean by top-down programming. You make procrastination a productive part of the problem-solving process by not losing yourself in the fine details until later.

  I remembered the bicycle museum in Belgium, where the technology development was displayed innovation by innovation. "Ainsi naquit la bicyclette": "Thus was born the bicycle," the label on the first example from the eighteenth century said. The men in top hats straddled steel two-wheelers in the diagram. There were no pedals, no brakes, no tires. In effect it was a running aid. So it answered the wishful question, Wouldn't it be nice if, when I ran, I weren't limited by the length of my legs, but rather could give the earth a push at regular intervals and compete with longerlegged people equally?

  The second series had tires, which must have answered the wishful question, Wouldn't it be nice if my push-bike didn't hurt so much? followed by the penny farthing, which answered the question, Isn't it sort of stupid to be pushing off the ground when I can crank a wheel like all the big mechanical machines do?, followed by the two-wheeled, modern-day "safety bicycle," which answered the question, Wouldn't it be nice if I didn't have to crack my head open every time I fell off my penny farthing?

  More Siebert lecture. "Now, in LISP, as in any language, we deal with primitives, means of combination, and means of abstraction, and also with idioms, or common patterns of usage."

  LISP is a slow language, but once you know it, it's very easy to read and understand a program because the programmer's time is the expensive part and the electricity needed to fire the logic gates open and closed in the chips is relatively cheap.

  "The key with large systems is communication," Siebert said.

  You want to communicate with other members of your group now quickly and efficiently so you don't waste time explaining messy logic to someone else. Besides, the someone else may be you later on-after the weekend when you come back to the problem-and if your code isn't clear you'll spin your wheels trying just to get back to where you were on Friday.

  "But we could talk all day," he said. "The machines in the computer lab are here to smarten you up."

  The problem sets help you achieve mastery, and sitting at the machines trying to wrap your mind around the concepts presented is the best thing you can do to attain mastery. When you are at the machine, the machine is your teacher, your dictionary; the machine is your thesaurus; the machine is your version of E. B. White's The Elements of Style. Why? Because anything that is bad grammar or bad spelling or bad usage simply does not work, and you have to try again and again until it does work
. Siebert said that there would be some tutors in the lab to help us if we really get stuck. They were sitting at the back of the classroom.

  In the lab. The computers were Chipmunks, donated to MIT by Bill Hewlett ('36)-Packard. These things were sort of megapersonal computers, each worth $50,000 if it were commercially available. But, of course, these costs mean nothing because they are coming down by a factor of 2 every couple of years. They were all lined up in a big room in the EG&G building, Building 34. I don't know why they called them Chipmunks, unless it was a jab at the users of the machines, the chip monks. None of them was named Alvin, so that wasn't it.

  There were fifty machines and not a lot of them were occupied so I sat down and tried to figure out how to log on. It was a little trickier than the Apple in Greene's computer room, because (1) there was no floppy disk to enter, and (2) the machine was already on. There was nothing on the screen, not even a little flashing rectangular cursor.

  I thumbed through the guide to the machine, looking for an instruction that said, "Press any key to start" or "Type 'ENTER' to start." There were no clues anywhere. It wasn't written on the blackboard. It is so obvious, such common knowledge, such an idiotproof concept that everyone should know intuitively how to start.

  After half an hour of reading and rereading the manual and looking through my notes, I asked the nineteen-year-old teaching assistant. Yes, it is humiliating to ask a mealymouthed sophomore for help, but age has nothing to do with ability.

  "Hit the space bar," he answered.

  Golly. I should have known that. The problem of the day was, Define a procedure (LISP for "write a program") that takes three (3) numbers as arguments and returns the sum of squares of the two (2) larger numbers. The problem was in the first chapter of the notes so it was by definition easy. No, it wasn't.

  How to proceed? I stared at the little bar cursor on the black and white screen that flashed annoyingly, as if to say, "Come on, slowpoke, you're boring me."