Jonathan Schaeffer

ONE JUMP AHEAD, One Giant Leap for Schaeffer

reviewed by Dap Hartman

This review should really be one line long: "Run - don't walk - to your nearest bookstore and buy this book". That is how, many years ago, I heard Leonard Bernstein describe L'Histoire du Soldat by Stravinsky. I did not know that piece then, but he was absolutely right. And so am I. Bernstein figured that no one would literally do what he had just said, and so he contained to elaborate on his recommendation. And so will I.

"By the way, did I ever tell you how much I hate writing?", Jonathan Schaeffer says on page 339 of One Jump Ahead, the book he wrote about his quest of becoming Checkers World Champion with his program CHINOOK. Gee, who would have guessed; the book is 496 pages long. But after finishing it, you will be left craving for more. It is clearly a case of 'the book that everyone has in him/her'. Schaeffer explains that he had to write it in order to exorcise some of his nightmares. At first glance it is just the story of how he headed a team effort to create a computer program that could beat the human Checkers World Champion. But merely a handful of pages into the book, you realize that it is much more that than. One Jump Ahead is a very personal account of what drives a man to undertake a quest like that. It deals with motivation, enthusiasm, and euphoria, all of which are counterbalanced by disappointment, frustration, and anger. Obsession, by Jonathan Schaeffer.

"The craving never goes away; it just diminishes with time". A quote from Trainspotting? No, from that same page 339. One of many moments of introspection Schaeffer confesses to. "[..] we went to Madrid to watch the World Computer Chess Championship [...] it was very hard to watch an event as a spectator when I was used to being a participant. [...] I guess I was just undergoing withdrawal symptoms". What is all this about 'craving' and 'withdrawal symptoms'? Well, as a reader of the ICCA Journal, you probably know what it means. Chances are that you too are 'infected' or 'possessed' in more or less the same way Jonathan Schaeffer is. Although, after reading this book anyone will agree that 'more' is hard to imagine. It is about power and control; it is you and the computer, and if you can tell it what to do, then you can accomplish anything you want. Become World Champion by 'teaching' (programming) your brainchild to be better than you will ever be. It is Searching for Bobby Fischer all over again. The sequel, if you please.

So why is this book about checkers then, and not about chess? Because even though Schaeffer worked on computer chess for more than a decade, he never reached the very summit there. That peak was only scaled last spring, when the entire weight of IBM was thrown behind the effort to beating Garry Kasparov. However, DEEP BLUE was not the first computer to beat a reigning World Champion. First, there was Hans Berliner's backgammon program which beat the human World Champion two decades ago. A big upset, but by everyone's account an incredible stroke of luck. Besides, it was only an exhibition match. Then what about the Kasparov and DEEP BLUE matches? Indeed, these were exhibition matches as well. Garry Kasparov did not put his World Championship at stake. Even if he had wanted to, he could not have done that. A major criticism against these two matches is just the: DEEP BLUE never earned the right to play Kasparov. Not many people believe that the machine could have passed the primaries. It is such a Narrow Path, as Jan Timman described his experiences of trying to be a World Championship contender. Am I drifting from the subject? I am sorry-let me return to Jonathan Schaeffer and his checkers-playing program, CHINOOK. You see, they did earn the right to challenge the World Champion. By starting at the bottom of the scale, CHINOOK gradually gained acceptance as well as respect in the checkers community. After the program earned the right to play Tinsley for the World Championship, the American Checker Federation refused to sanction that match. At that point, Marion Tinsley did something incredible: he resigned his title in order to play CHINOOK in a match. When Kasparov had his major disagreement with FIDE, he did not resign the title; he started his own federation. Reading Schaeffer's account quickly makes you aware that checkers is a whole different game.

One Jump Ahead pretty much described Schaeffer's life between 1989 (when the idea to write a checkers playing program first came up) and 1995 (when CHINOOK was retired). He makes the reader (as well as himself) painfully aware that what seemed to be a relatively simple task-"in only two months time we already had a program that was good enough to beat a former Canadian Champion. Surely it wouldn't be that big a leap to improve the program to the level of the World Champion. [...] This was going to be easy" - turned into an obsession which consumed the better part of six full years of his life.

Schaeffer obviously borrowed the basic format for his book from Clifford Stoll's The Cuckoo's Egg, wherein a narrative style is interspersed with sections from a logbook. At times, it reeks of old Star Trek episodes. You can just imagine Schaeffer sitting the captain's chair, when he scribbles this thoughts into the logbook: "Programmer's log, CHINOOK project day 442. Friday, August 18, 1990. On the 442nd day since our quest for the World Championship began, the human World Champion sits down to play the mighty CHINOOK". All shields are up, as we have just spotted a Klingon battleship in this sector' - I cannot help getting flashbacks like that. But there does exist a logical connection: it is definitely where no computer has gone before. By using the present tense in the logbook, Schaeffer creates a nice change of pace, which actually makes the events described happen right here and now. Everyone who has operated his own computer program in a serious game will relate to what Schaeffer describes. The anxiety before the game, maybe because a bug was fixed, or maybe because a special line of play was prepared, asking yourself: 'How well will we do this time?' And the mood swings when the evaluation values change as the game progresses. It is hard to believe Schaeffer's statement that he would generally read a book while a game was in progress. it must be during the (many) boring games that occur in checkers, because the annotated games in this book are all interesting for one reason or another. Some because of a brilliant move, others because of a bug, or because CHINOOK sees the outcome of the game after merely a handful of moves has been played.

One of the great achievements of this book is that it portrays Marion Tinsley as the greatest checkers player who ever lived. Maybe even the greatest player of any game, ever. Schaeffer's admiration for Tinsley is formidable, and very infectious. He never passes up on an opportunity to praise him, both as a human being:"[...] a sportsman and a gentleman in the truest sense of the word", as a checkers player: "I am watching a great man in action, and all I feel is tremendous awe and respect for him". Tinsley's record as a player is unbelievable: In the 45 (!!) years that Schaeffer calls his 'peak years', Tinsley lost only 5 (!!) games out of the thousands of games that he played. There is no equivalent achievement in any other game or sport. It is as close to perfection as humanly possible (the title of Chapter 8). Analyzing a huge database of games that Tinsley played in the past, CHINOOK found a few instances where he may have made an error. As it turns out, the program did not calculate deep enough to see that there was no error: Tinsley was right. If you have never heard of Marion Tinsley, this book will convince that he truly was the greatest player the game has ever seen.

Thus a win against Tinsley is very, very rare. Most of his human opponents would be more than happy with a draw, and that attitude has spoiled much of the fun for him. With no one willing to take a little risk (which is generally what you have to do to try and win a game) the arrival on the scene of a fearless computer was a very welcome and refreshing experience for Tinsley. "I feel like a teenager again" (the title of Chapter 10) said the 63-year old Tinsley after playing CHINOOK for the first time. The program is not impressed or intimidated by it opponent, but the programmer is. Schaeffer behaved just like a human checkers player when he celebrated its drawing Tinsley as a victory: "We did it!!! Four draws against the reigning World Champion!". that was in August, 1990. It would be another two years (almost to the day) before the computer's evaluation leaps up on move 21 of the eighth game that CHINOOK and Tinsley are playing for the World Championship: "That's it! A gigantic score of 286 means that CHINOOK is winning everything in sight [...] we're going to win. Win! Win! Win!". The enthusiasm jumps of the page, and grabs you by the throat. Two years, amounting to 130 pages in the book, separate these two milestones, and the reader has been privy to most of Schaeffer's ups and downs in the meantime. At this point in the story, Schaeffer has really convinced the reader what a phenomenal accomplishment it is to have CHINOOK be only the fourth player in over 40 years to win a game against Tinsley. And with his style of writing he is also able to convey to the reader the inner feelings that accompany this accomplishment.

"Am I still a jerk?", Jonathan Schaeffer wrote as a dedication in my copy of the One Jump Ahead. No. Never have been either. But if I had not known Jonathan before I read his book, I certainly might have gotten that impression. In his eagerness to show the reader how honest he is, Schaeffer frequently mentions broken promises to his wife, and tries very hard to make us believe he is inflicted with an addiction that is responsible for his behavior. Off to a good start (on page 16), Schaeffer writes in the 'Programmer's log' for Day 1 of what was then called the 'checkers project': "[according to my wife] I am dishonest, insensitive, self-centered, and, to be quite blunt, a jerk". The sad truth is that I honestly believe that Schaeffer could not have accomplished what he has in any other way. He is certainly not a nerd, but he is addicted to excel in everything he does. It just happens to be related to computers, most of the time. He justifies his behavior (to his wife, to the reader, and also to himself) by arguing that this is the once-in-a-lifetime chance to attain his life-long dream, and he promises to make up for it as soon as it is over. After reading the book you know that it was never really over because there were many such chances, and many such dreams, including writing One Jump Ahead. The sad part is to realize that there must have been many such promises. "Oops, a roadblock already", described Schaeffer his gut reaction when his wife says they should relax and take a holiday. And it is only Day 1 of what will amount to a 2,134-day project. Jonathan Schaeffer, family man.

There is one thing in this otherwise fabulous book that I disagree with. To make the checkers games easier to follow, Schaeffer uses the algebraic chess notation instead of the numbered squares which is customary in the checkers literature. Schaeffer is probably right in assuming that more (computer) chess players will read his book than people from the checkers world. Nevertheless, I feel he should have used the standard checkers notation. You do not translate Don Giovanni into English just because most Americans do not speak Italian.

What I like best about One Jump Ahead, is that it is a very personal story. The subject is almost irrelevant; it could have dealt with his quest to become the fastest man on the 100-meter hurdles. What matters is the man behind the story. Even though there was a whole team involved in the project, everything revolved around Schaeffer. He is very open about his feelings and he tells you exactly what he thinks and feels, even though that might possibly offend certain people. Especially when he confesses which people he likes and which he does not like at all. Schaeffer's style of writing is very informal, and his words directly reflect his feelings, whether he is excited: "Was he [David Levy] calling about, please, the Tinsley-CHINOOK match? Yes, yes YES! David said he had money for the match- not a lot, but enough", or frustrated: "I can't believe how quickly the match has turned on me. How could CHINOOK be so dominating at the start of the match and then stumble so badly, as if overconfident? The program is behaving as if it had a human weakness. No, no, NO! I'm the overconfident one, not the computer". He makes you relive some of his own edge-of-the-seat-thrilling moments: "But wait! Something is wrong. Chinook is printing out illegal moves. Damn! A surge of panic and helplessness overcomes me. Are we going to forfeit this first game? This is my worst nightmare come true; what a horrible way to lose a game. We'll never be able to live down this embarrassment. No! CHINOOK is analyzing the wrong position!" It makes the book a very enjoyable fast-paced read which I found almost impossible to put down. The last 60 pages were the hardest for Schaeffer to write, and he admits that it was a very painful but hopefully therapeutic exercise. What started as a joyful enterprise ended in human tragedy with a lot of bad feelings and innuendo. The only ones not (yet) affected by the final four chapters will be computers.

"Playing chess is like looking out over a limitless ocean; playing checkers is like looking into a bottomless well", was how Marion Tinsley distinguished the two games. 'Chess is an ocean while checkers is merely a well', reflects most people's attitude towards these games. Tinsley's description is very appropriate because it is obvious to even a casual observer that the ocean appears to be limitless. But it needs a closer inspection of the well to find out that it is bottomless. Schaeffer himself gradually gained respect for the game of checkers ("After all, it was going to be easy"). But as with most analogies, Tinsley's description too is not 100% correct: there is a bottom, and someday the bottom will be reached. Just like Jacques Piccard eventually hit the bottom of the deepest trough in the ocean, someday the game of checkers will be solved. I wish it to be Jonathan Schaeffer to do that, but it suggests yet another once-in-a-lifetime opportunity.

[University of Alberta] 
University of Alberta 
[Department of Computing Science] 
Computing Science