31 July 2017

The First USCF Rating System

In last week's post, The First USCF Rating List, I reproduced two Chess Life (CL) articles introducing the first USCF rating system, along with the first page of the rating list. That first list, published in the 20 November 1950 issue of CL, was accompanied by a more detailed explanation of how ratings were calculated. The article was titled,

National Rating System by William M. Byland
USCF Vice President in Charge of Rating Statistics

The first of Byland's articles said,

In this series of [four] articles we will attempt to explain the operation of the National Rating System adopted by the USCF Board of Directors at Detroit last July. In selecting a rating system to fit the needs of American chess players, we have been guided by four basic principles:

1. Universality • Our system is universal in its application, and covers all types of competition: national, regional, state, city, and club tournaments. At present, only tournament competition entitles a player to a rating. We are working on plans to include team and individual matches, and hope to have this phase of the system in effect in 1951.

2. Mathematical Operation • Our system is completely mathematical in operation, without bias or prejudice, and its mathematical correctness has been attested by several leading actuaries in the United States and Canada. Because the system registers a player's failures as well as his successes, we now have an accurate yardstick for determining the relative playing strength of United States players, based not on reputation or self-claim, but upon cold performance facts.

3. No Barriers to Progress • Our system has no artificial harriers to impede a player's progress. It is based on the principle that only his actual performance record should determine his classification, and no bars are set up to prevent a player's rapid progress from being reflected in his current rating -- he is not required to progress laboriously upward from class to class.

4. No Premium for Inactivity • Our system, on the contrary, encourages activity on the part of all players, but does not, thereby, render tournament participation a hardship. In order to be rated, a player must participate in at least one rated tournament every three years.

Rating lists will be published twice each year: as of July 31 and December 31. Our first rating list, as of July 31, 1950, appears in this issue, and covers 2306 players; on future lists, only ratings of USCF members below the Master class will be published, and the names of inactive players (those who have not participated in a rated tournament for three years) will be omitted. Also in this issue is a listing of the 582 tournaments, covering a 30-year period, used in determining the current ratings of the players on the list. This tournament roll makes no pretensions to completeness, and contains only those tournaments whose cross-tables of play were published and readily available. It is interesting to note how the yearly list of tournaments has expanded since 1921. and the increasing publicity these events have received over the years (in which field CHESS LIFE has been an undoubted pioneer). For the long labor of compilation and computation involved in these listings. which furnish an invaluable base for future ratings, we are deeply indebted to Rating Statistician Kenneth Harkness.

(To be continued)

The article preceded a list of the tournaments included in the rating calculation, which dated back almost 30 years. The start of the list is shown below, where the last column is an average rating for the event.


[...]

Because they take too much space for a blog post, I won't reproduce the other three articles in the Byland series. They covered the following topics:-

CL, 5 December 1950
* How a Player's Average Rating is Computed
* How Performance Ratings are Computed

CL, 20 December 1950
* Computation of Performance Ratings for Round Robin Tournaments
* Computation of Performance Ratings for Swiss System Tournaments

CL, 5 January 1951
* Special Provisions of Rating System

That fourth article (5 January 1951) was accompanied by a CL editorial.

THE RATING SYSTEM

Enough has been written by hasty as well as thoughtful critics to indicate that a good many features of the new National Rating System have not been properly understood nor correctly evaluated. We have therefore asked Mr. Byland to prepare an article for an early issue in which the more important points of misconception can be stated and clarified. But it might not be amiss at this time to repeat a few salient points without awaiting for Mr. Byland's more complete statement.

First. because of the fact that not all tournament scores were available in sufficient detail for analysis, certain players (particularly in certain sections of the country) suffered from some injustice in the compilation of their initial performance ratings. We are aware of this fact, which will be self-remedying in time as more recent tournament reports are received in fuller detail from more tournaments. It was a fault that could not he removed from the first compilation.

Second, it must be remembered that these performance ratings do not pretend a permanence for all time, nor do they evaluate a player's total record over the years. Unlike FIDE master titles, these ratings are based solely upon recent performance and in no case represent an honorary degree for past performance. Therefore, there are a number of players whose best years of tournament performance occurred before the period covered by the ratings. Their present standing. therefore, does not indicate (nor can it) the exalted position they would have held if this system had been in operation some twenty or thirty years ago. For example. Dr. Edward Lasker's present performance rating is a very modest one compared with what it would have been if these ratings were based upon performances some thirty years ago.

Third, it must also be remembered that these ratings are based exclusively upon performance in American events. For that reason, such outstanding players as USCF Vice President Hans Kmoch and UCSF Life Director George Koltanowski are excluded. Their notable performances abroad have no bearing upon performance ratings in a national system.

Fourth. the national ratings do not pretend to evaluate ability or potential talent comparatively -- they merely record results of actual performance mathematically as a convenient yardstick to settle a number of disputes as to precedence. Such a standard is essential (even in chess heavens like the USSR which has a very elaborate system of rating) and are common to a number of other sports besides chess.

Finally, let us repeat once again since a number of readers seem to have misinterpreted previous statements Any tournament with two USCF members in the entry which is five or more rounds and not a speed or restricted move tournament is eligible for rating. BUT future rating lists will only contain the names of USCF members. The USCF will of necessity keep record of all ratings, but we will publish only those of members in good standing whose dues are supporting the cost of such an elaborate and exhaustive system.

Montgomery Major

Although the theory behind the rating calculations has evolved through the years, and the name of the organization has changed from USCF to USchess, the rating system remains as the nucleus of the services offered by the U.S. chess federation.

No comments: