Reprinted with permission of Bill Williams, Badger Chess editor from the June 1982 edition.
An Interview with Arpad Elo
Badger Chess Assistant Editor Ted Babcock interviewed Professor Arpad Elo, “father” of our rating system during this year’s Elo Open in Madison, April 2-4. Babcock described Elo as being “most gracious, anticipating many of my questions, and giving answers in great detail. . . an enjoyable experience for me.” This is part one of the Elo interview, part two will appear in the July issue.
When and where were you born?
Well, I was born in Hungary on August 25, 1903. A long time ago.
And when did you move to the United States?
In 1913. I just escaped World War I; that is our family did. That was lucky. We moved into the Cleveland area, where I went to high school. In 1921 I went on to the University of Chicago and played chess there in the Chicago Chess League. Chicago was a good chess center, a good many strong players. Then in 1926 I took a position at Marquette University in Milwaukee and, except for a break during the war years, I was there until my retirement in 1969.
Who were some of the better players in Chicago at that time?
There was Samuel Factor, who won the Western Championship a number of times. He was a strong master. And Herbert Albohm. Those were the outstanding players.
Edward Lasker lived in Chicago for a while; although I have never played against him—I did against the others.
For many years we had an annual match between Milwaukee and Chicago. It was a pleasant, generally late in the summer.
What was your playing strength at the time?
I would say high expert, maybe a minor master occasionally. The first USCF ratings during the Harkness era put me at about 2230, but it’s difficult to get the correlation between the present day and that day. One would have to carry it through historically from one era to the other. When I developed the Elo rating system, I retained the general features of the scale so people would not be confused by changes of numbers. Of course, I gave meaning to the differences, which were more or less arbitrary up to that point. The statistical basis of my system are completely outlined and developed in my book (The Rating of Chessplayers, Past & Present –Ed.). There is a full history which should be known by now.
When did you devise your system?
I was appointed in 1959 by the late USCF president Jerry Spann to overhaul the system and I realized then that this was essentially a measurement problem involving statistical methods, so I started literally from scratch with the basic probability theory to lay the foundations of the system. The numerical values of the ratings are arbitrary. I retained the four digit numbers, not because the ratings are accurate to that many significant figures but so that people would not be confused by the changover. The 2000 level was selected as the level of a strong club player. Actually, it might have made more sense to put it at the 1000 level, but the low ratings would be so close to zero that ballast numbers were added to make the difference seem, at least psychologically, less. Then with the definition of the class intervals at 200 points, the other ratings were automatically generated by the system.
Sometimes I think I’ve created a Frankenstein’s monster with this system because some of the young players become just like race track habitues who never really see a race; all they do is peruse the tote sheets.
How long had the Harkness system been in use before your system replaced it?
The Harkness system was started in the early 50’s, so it was in use about eight years—but in that time it developed such unrealistic numbers and deflation that something had to be done. It had two formulae for calculating performances: one, the short formula, was for tournaments of nine rounds or less and the other was for ten or more rounds. It was inherently a deflating system because with the short formula—and most of the weekend tournaments were five rounds—it was possible for a player to score a perfect 100% and, if his competition was weak, lose rating points. The deficiencies became obvious.
The thing was, there was no theoretical basis for the Harkness system where you could examine the system’s validity. In my book I’ve given the theory to the Harkness system and you can see that it’s theoretically untenable because the implications are that all performances that a player can have are equally probable, which is just not the case. Normally a player will perform closer to his average more frequently than he will at the extremes.
The adoption of the (new) system was by no means a smooth thing. People resist change in general, you know—it’s human nature—and the people who did the computations in the office were comfortable in what they were doing. When it required a change and mastery of a new system they were like the mythical Marine sergeant who was unfrocked and placed into the regular army—He spent most of his time memorizing his new serial number.
Who was pushing most for the change? Did it come from the players, or--?
Yes, demand for change was from the players and fortunately Jerry Spann, who was really a class guy, saw the importance of having a system in which people have confidence. He supported me, and eventually we got the adoption. Everything has not been rosy there either, because of changing circumstances in the federation. With the new rating system followed by the Fischer boom there was a big increase in the membership. The influx of so many new people into the system created special handling. I don’t know whether they are solved yet or not. But, of course I’ve been out of it now for five years, so I haven’t followed all the ramifications of the changes they’ve made since. I’ve been too busy with the FIDE side of it.
The FIDE application, I think, in many respects was the most satisfying and the smoothest operation because there I had more or less complete control. Up until 1980 I did all the calculations myself, and there were not so many that I couldn’t do it with a desk calculator. I have a Hewlett-Packard 97 calculator that is programmable, so I did all the ratings and could see what to do about any problems that developed. I had the full cooperation of the Qualifications committee. The static I got was from certain quarters of the U.S.
What was that all about?
Well, I’d rather not get into that because it would get into personalities. It’s become a personal matter between myself and certain others in the United States. Because I did not go along with attempts to corrupt the system in certain ways I was vilified and accused of all sorts of misdeeds—that I was discriminating against American players and that sort of nonsense. Fortunately, the people in FIDE knew the score too, and what was being done.
Has anyone complained about the list of top players you compiled in your book; that some players should have been higher or lower ranked?
Not really seriously. Of course every great player of the past has his circle of fans, so it has become a subjective thing.
You think Alekhine was stronger than Capablanca—well, he was at the time they played the match, but when you see their overall record, Capablanca had a plus score against Alekhine and they were close enough in age so that was not a great factor. That was the single case that was called to my attention—some people thought Alekhine was stronger than Capablanca. Capablanca had a plus score against Lasker, and he had a plus score against both of those players, even against Lasker despite his age advantage.
Of course there are always the Morphy lovers and the Morphy detractors. But people don’t realize that when you apply such a statistical system which evaluates players, you must consider the place of the players in their own milieu. You can’t compare a Morphy of 1857 to a player of 1957 because the player of 1957 has so much more accumulated knowledge to build his play on. You have to consider each of the historical masters in relation to the players of his own time, with an equal amount of accumulated knowledge available.
You know, even as an expert player I need to get into games with players rated below me and in the opening they would play like masters. But then you get them into the middlegame or endgame and they would fall apart. Then you realize that what they were playing is some variation that they had seen someplace and prepared. How can you say that that player is inherently better because he had done that? It’s when he’s on his own resources that a player’s strength really shows, and that comes in the middlegame and endgame.
So, when I see such comments as that Morphy could be beaten by any master or expert of the present time—It’s just not so. Maybe, if that expert had all the current opening analysis available and Morphy didn’t—But the thing is that Morphy was the innovator; he laid the foundations of master play and then others after him, Steinitz and Nimzovich, codified that master play. It is true that of all the games Morphy played, I don’t think that more than 60 or 70 were against first class opposition. He played matches with Paulsen, Harrwitz, Anderson and Lowenthal—that’s all. Any modern master will play 60 or 70 or 100 games against master competition in any one year. Morphy in his lifetime played only 40 games against Grandmasters. So some of the criticism may be valid, but at the same time you have to look at the creativeness of the man and at the innovations that he brought about. There can be a lot of argument and discussion on the validity of the historical ratings, but that’s only a small part of the book. The book is really written more for posterity than for the ideas in there, which are beyond chess application.
Did you ever try to calculate your own peak rating?
Yes, I was about 2230 about 1935-45, after I had two draws with Rueben Fine, a draw with Arthur Dake, and a few other successes against minor masters.
I played in the very first US Open in 1934. It was held in Chicago. The following year the second was held in Milwaukee—actually we called it the American Chess Federation Championship. The ACF was organized because of a general dissatisfaction with the National Chess Federation, which had become kind of a moribund organization. It (the NCF) was a rich man’s club centered in Chicago. In 1946 the two organizations merged.
So in 1935 the 2nd ACF was played in Milwaukee. Reuben Fine played. (Isaac) Kashdan, Arthur Dake, and Albert Simonson a very promising player from New York, were there, too. The tournament was run in sections, with the top three from each section going on to the master section. I was lucky enough to qualify. There were 32 in all—nothing like nowadays when we run into the hundreds. That was the first time I met Reuben Fine, and I managed to draw a game I should have won; actually he should have annihilated me in the first place, but he misplayed it, and then I misplayed it—a comedy of errors. It ended in a draw while I was two pawns up in a Queen ending, but inadvertently allowed the position to be repeated three times and he claimed a draw. We became good friends. I met him again in Dallas, in 1941 I think. We played twice in that one: the first game ended in a draw and he won the second.
There are books of those tournaments that should be available in the Milwaukee Public Library.
Do you have a favorite player?
No, not really. I’ve always been interested in the historical and cultural development of chess, its place in civilization. I made a great deal of study of that, but I cannot say that any one particular player is my favorite. I’ve made many good friends, so there are a many players whom I admire a great deal. I think of Reuben Fine—maybe some people didn’t like his personality, but he was a very erudite person whose interests were very wide. And (I’m speaking now only of the players I know personally), Arnold Denker was a charming person; he’s so effervescent, and even at his age—he must be pushing 70 now—he’s so full of energy and enthusiasm that you can’t help but admire and like the guy.
Have you met Reshevsky?
Oh, yes. I met him first in that Chicago tournament in 1934, where he beat me. The next time was in a simultaneous in Milwaukee where I beat him. He remembered that the second time around because he bore down and said that he “had to remember that!” Reshevsky hated to lose in simultaneous exhibitions. It was comical in a way.
Did he beat you the second time?
The second time he beat me. He concentrated!
Have you played in many exhibitions against leading players?
Yes, you’d be surprised that I played against Maroczy when I was a student at the University of Chicago. I played against Frank Marshall—these are the real old-timers, you see—and Kashdan, whom I beat, and Fine, Reshevsky, and Fischer. Of course I lost to Fischer, but I also played him over the board in one of those Milwaukee tournaments (He is referring to the Western and North Central Opens—Ed) when he was still a young man. I lost to him in a game that was a dead draw. I had a hallucination about the position—but of course no chess player ever loses a game, he throws it away, so I can say that I threw the game away—but it was a draw. In an ending with two pawns each--his were connected and mine isolated—with bishops of the same color, I made just one mistake trying to advance the pawns and he pounced on it immediately.
END OF PART ONE. PART TWO IN JULY