The Courier-Journal from Louisville, Kentucky · Page 212

"Okay," Ken says. He looks at he screen, and so do I. What is printed there looks like this: RNBQKBNR PPPPPPPP P FRANKFORT, KY. 11 A HAT everybody tries lf when they play chess V V with this computer," Ken Presley says, "is using some crazy, far-out set of opening moves. They think they'll confuse the thing." I nod sagely, looking at Ken's youthful, chubby, lightly bearded face as though I have always understood the silliness of trying to confuse a chess-playing computer. Between us sits what looks like a small portable TV set attached to a typewriter; on the face of the TV screen rows of white numerals silently appear, hop about, disappear, and are replaced by letters. It begins to dawn on me that the letters are the initials of the pieces in a chess game: P, B, Q, R, N, K. Is the damn thing practicing? Tuning up? Ken is very intense; his stubby fingers tremble slightly as they rest near the keyboard of the machine. A Louisvillian, he majors in computer science at the University of Louisville and has spent about two years developing the program for chess playing that I am about to try my hand at beating. I look down and notice that my fingers are trembling too. It's only a dumb machine, I tell myself. Like pinball. But I almost never could beat pinball machines either. We are waiting here now, on the ground floor of the Capitol Plaza in Frankfort, for this number-and-letter-flashing console between us to be properly hooked up by telephone line to the actual computer in the Speed Building at UL. Something about the lines isn't working, probably something to do with the recent tornadoes. I had seen telephone linemen at work, in the devastated town of Jett, a few miles away, on my way here from Ohio. "You can't confuse the program," Ken goes on. "You might beat it by seeing more moves ahead than it can see or four half-moves but it will come up with a good move no matter how crazily you open." I nod my head. "Sure." What I don't tell Ken is that I sat up with a "Why not let it think about all of them? It thinks fast enough." "No it doesn't." Dana smiles shyly. "No computer thinks that fast. The number of possible chess games that could develop from any given move is probably as great as the number of electrons in the universe." "Oh," I say. My mind has abruptly switched from pictures of forests of game trees elms with Monopoly sets for leaves? to contemplating the vastness of space and the plethora of busy atoms that inhabit it. "So you have to program the computer to prune the tree?" "Right. First we plant the game tree and then we have the program set to prune forwards or backwards." "And does the UK program do this the way the UL program does it?" I am beginning to feel better now, since I'm always more confident dealing with people than with machines. "Not exactly," Ken breaks in. He is toying with a chess piece. There is a board set up between me and the console, and - Ken ' will move the white pieces as the computer tells him to, and input as they say my moves into the machine. I don't know exactly how I have wound up with the fractional disadvantage of playing the black pieces and thus being the second one to move but I suspect there is a sublimated hustler beneath the scientist in this burly young man. I am pretty certain by this time that he wants his program to beat me. Badly. To humiliate me, in fact. "But the difference in pruning techniques is part of our seer ts. The ones we keep from one another before tomorrow's match," he says. He sets the chess-piece, a knight, back on its home square. In the room are several rows of black plastic upholstered theater seats for tomorrow's audience mostly empty now. In one of these is sitting Katie Nooning, who has been one of the main people in charge of this project from the beginning. Katie is thin, young, a bit owlish in her big, horn-rimmed glasses. She seems far too gentle and feminine to be an instructor of computer science at UK. "I think everything's ready now," she says. chess-playing friend two nights before and practiced the most outre opening we could find, at the very back of "Modern Chess Openings," the Bible of the subject. It was called Durkin's Attack, and started with knight to queen rook three. The idea had been to confuse the computer. "The computer patrols the squares," Ken says, "trying out every possible move with every piece. Then it considers what will happen for every important reply you might make. Then it patrols that situation. And so on." I am beginning to feel a tinge of panic. could never "patrol the squares." But then I couldn't multiply 27-digit numbers together correctly in a microsecond. I try to look intelligent and to ignore the quiet little light show on the face of the console, although I am beginning to think I would rather face Bobby Fischer's steely, winner's eyes than that outward and visible sign of God knows what inward, chthonic presence. "How can the machine know which of my possible replies are important ones?" I ask. AcROSS the room, at another, different kind of console one that can print out chess moves on a sheet of paper instead of a screen has been sitting a lean, amiable young man with long hair. Dana Ross of Paducah, a University of Kentucky graduate student in computer science, is captain of a four-man team that has programmed a UK computer for a match with UL that is to be played here machine against machine the following day. He speaks up, quietly and pleasantly. "Essentially," he says, "the program uses what we call a 'game tree.' The branches are all of the possibilities that ramify from given moves. What we have to do is program the computer to prune the tree. It's very tricky to do." "I can imagine," I say. "You use the alpha-beta routine," Dana says, nonchalantly, as though he were discussing brands of beer. "Which is a form of backward pruning towards the bottom of the tree. And of course there's forward pruning. Both of these limit the number of possible games the computer has to think . about." PPPP PPP RNBQKBNR It is obviously a representation of the set-up of a chessboard at the " opening of a game, right after White has moved pawn to king four. So the machine has made its first move. I am somewhat comforted by what it has chosen P-K4 since it is pleasantly ordinary, about what you would expect the guy next door to open with. But then Bobby Fischer uses almost nothing else. Below its little mock-up of a board the machine has printed: "INPUT YOUR MOVE." Ken moves the white pawn to king four and looks up at me. I reach out and move my queen bishop pawn to the fourth rank, starting the Sicilian Defense, a fairly routine reply to White's opening. Ken taps away at the keys on the machine for a minute, telling it my move. Then he sits back, folds his hands across his stomach, and watches the numbers appear and flicker on the soreen as the . machine begins thinking. Or, I guess, patrolling. "How long will it take?" I ask, expecting him to say something like seven seconds. "A little over four minutes." "Four minutes?" "Yeah. Something went wrong and we weren't able to hook into the . big computer here in Frankfort. The reason we came here in the first place was partly so we could tie in with that biggie. It would have been a lot faster. The Hewlitt-Packard machine we are plugged into will make the same moves because the program is the same but it'll just take longer to make them." There goes another of my preconceptions. I had expected to be at a kind of psychological disadvantage, sweating out my moves for long minutes while the machine would contemptuously be flicking its moves out in microseconds. It turns Continued WAITER TEVIS, professor of English at Ohio University in Athens, Ohio, is a groduote of the University of Kentucky ond the author of two novels, "The Hustler" and "The Man Who Fell to Earth." He has written a number of articles on the game of chess. SUNDAY, MAY 19, 1974 35