I outlined a series of eight articles written in 1952 by Kenneth Harkness. He added a ninth article in the 20 September 1952 edition of Chess Life (CL).
Swiss System Pairings
by Kenneth Harkness
USCF Rating Statistician
The pairings of a Swiss system tournament produce some peculiar results, as anyone who has played in those results knows well. The winner's title may be clouded because he failed to meet some of his strongest competitors. Others place high in the final standings after meeting comparatively weak opposition. A player may shoot up from nowhere in the last round or two and outdistance contestants who played far stronger opponents.
In a tournament for an important title, the Swiss System must be regarded as inferior to a round-robin if the winner does not meet all the strong contenders. However, the Swiss has a great many practical advantages. These advantages so greatly outweigh its known defects that the system is now used in practically all state, regional and national tournaments with the exception of the United States Championship. If a better method of pairing contestants will cure the faults of the Swiss System, the quality of all the present tournaments will be improved and the system can he used for the U.S. Championship itself.
As an example of what can happen, we present in the table below an analysis of the pairings for the top twenty players in this year's U.S. Open Championship at Tampa. In doing so, we imply no criticism of the tournament director. Our quarrel is with the present method of pairing by lot, not with the director who follows standard procedure in this respect.
The table showed how many top players each top player faced.
Rank, Player, Score, Opponents Among Top 20, Opponents Below Top 20
Harkness continued with comments on the table. They provide more detail than is needed for this blog post, but I like the background about U.S. chess in the early 1950s.
Bearing in mind that the winner's pairings are the first consideration, we are bound to ask why Larry Evans played the men who came in 42nd, 47th and 49th instead of three of the strong contenders he did not meet -- especially Hearst. Mengarini and Donovan, three rated masters who performed well at Tampa. The answer is that Larry played the opponents who finished below the top twenty in the first three rounds of the tournament. With 76 players in the contest, the luck of the draw gave Larry three opponents who failed to make the grade later. Being the highest rated player by a wide margin, the U.S. Champion would probably have kept the open title in any case. Even if he had played Hearst, Mengarini and Donovan, Larry would probably have risen to the occasion and put forth the extra effort needed to win the tournament. However, the actual outcome cannot be considered entirely satisfactory. After all, Mengarini beat Reshevsky in the last U.S. Championship!
Below top place, it is clear that some of the men in the list might have finished lower if they had met stronger opponents. Our sympathy goes to Jimmy Sherwin who was unlucky enough to draw the strongest field of the entire tournament. Measured by the rating system, Sherwin's competition averaged 2306 points! Steiner also met pretty stiff opposition -- stronger than most of the players who finished above him. While Sherwin and Steiner were batting their brains out against practically every master and leading contender in the field, some of the other players coasted in ahead of them by scoring against comparatively weak opponents. Needless to say, the players who came in below the top twenty were not pushovers by any means. Many were probably stronger than some of the prize-winners who slipped into the money brackets on pairing flukes. However, all the active masters placed among the top twenty. and only a few of the strong experts failed.
It has occurred to this writer that the rating system might be used to advantage when pairing the contestants in a Swiss System tournament. Based on this conception, we have developed a method of pairing which may correct most of the faults and inequities described above. At present, the method is theoretical. It has not been tested in practice, so it remains in be seen whether the theory is sound With the co-operation of the directors of some forthcoming tournaments, we hope to check the results achieved and report the outcome later.
To use the method successfully, most of the players in a tournament must have national ratings. We hope the day will come soon when practically all players are rated, and we are rapidly reaching that goal. In the U.S. Open this year, only 5 of the 76 entries had no previous ratings. However, we cannot guarantee that this method will help much if you are running a tournament with a large number of unrated players. Furthermore, the method Will prove most effective when nearly all the entries have given us an opportunity to measure their ability by playing in several tournaments. A rating that is based on the results of only one or two tournaments is not necessarily a true indication of a player's strength.
Since the method is based on the rating system, the ranking of the entries must be done by your rating statistician who alone has all the necessary data. The up-to-date ratings of some players may be higher or lower than the published list indicates, and a great many names in our files may be missing from the list. If you wish to test this method, mail a list of all the possible entries, giving their full names, to this writer at the address given in the masthead of CHESS LIFE. We will send you by return mail the up-to-date ratings of players on your list. The provisional ratings of players who have competed in only one rated tournament will be marked with asterisks. Then, about an hour or two before the tournament begins. You may telegraph the full names of unexpected entries and we will wire back their ratings (collect!) adding the prefix "pro" to the name of a player with a provisional rating. For example, PROWILLIAMS 1850 Would mean that player Williams, has a provisional rating of 1850. Please note that all ratings supplied for the purpose of ranking tournament entries are confidential, for your own use exclusively as tournament director.
The pairing method is explained in the following paragraphs:
1. Make up a ranking list of all entries, arranged in the order of their ratings, from the highest down to the lowest. Add at the bottom the names of all unrated players, arranged in alphabetical order. [...]
The rest of the Harkness article, which nearly filled the equivalent of a full page in an issue of CL having only six pages, gave eight steps for making the pairings. These steps -- including advice about the 'fundamental rule' of Swiss system events ('a player must not meet the same opponent twice'), about color allocation, and about unrated players -- are well known to anyone who has played in a Swiss. What tournament was the first to use this method of pairing?
No comments:
Post a Comment