Original Published Ratings
by Mark Orr
created August 12, 2012; last updated July 24, 2014

In a Nutshell

For any given past rating list your official rating includes corrections for all known errors that occurred prior to the publication date, no matter how long after that date they were discovered. Your original rating only includes corrections discovered before the end of the month in which the list was published. A more detailed description follows.

Human Error

Part of the process of rating chess tournaments involves data collection and entry and is thus susceptible to human error. Such errors, when they occur, may not be detected until after the tournament has been rated or ratings lists have been published. Examples of the types of errors that can occur are the identification of players and the recording of game results.

Corrections

The new rating system, which began in January 2012, has the ability to easily correct mistakes once they are discovered, to rerate tournaments and to republish rating lists. However such corrections will typically lead to changes to many ratings (even if only by small amounts in most cases) because errors have an unfortunate tendency to propagate from the directly affected players to their future opponents and their opponents' opponents and so on.

Original Ratings

When a rating list is first published every player gets two ratings: an official rating and an original rating. Initially the two are identical. If any corrections are made to the list during the month of publication, both ratings are updated together and remain the same. However, when the month ends the original ratings are frozen and any subsequent corrections will only change the official rating. Thus players have the ability to tell if their rating has been affected by corrections performed after the month of publication by comparing the two versions.

Only official ratings are used for players' rating histories and for such things as WAR, Top Juniors and Top Improvers. By default, the main Ratings page only shows official ratings but original ratings can be revealed by clicking show original ratings.

Example

  • Player A plays a tournament in December 2011 but is misidentified as player B in the rating report
  • The January 2012 rating list is published and, unknowingly as yet, the ratings for both A and B are in error
  • A year passes, it is now January 2013
  • The mistake is discovered and corrected (the player in the December 2011 tournament is assigned A's ID not B's)
  • That tournament and all subsequent tournaments (over 100 of them) are rerated
  • The January, May and September 2012 rating lists are republished but their original ratings remain unchanged
  • The official ratings for A and B in those lists are now correct but different to their original ratings
  • The ratings of some of A and B's opponents after December 2011 may also have some differences, but smaller ones
  • Some of their opponents' opponents (etc) may also have changed ratings, but with even smaller differences
  • The January 2013 list is published and all official and original ratings are identical again, since both include the correction

Frequently Asked Questions

How are errors detected?

To a large extent that's down to you, dear reader! Given that rating calculations are exposed in full detail on this web site, any bugs have a good chance of being spotted by the people with the biggest incentive to squash them: the players themselves. If you do ever spot one, please email the tournament controller and rating officer with details.

Is it possible to obtain a detailed explanation for the difference between an official and original rating?

The system does keep records of corrections made and it may be possible to use these to explain differences which are the direct results of corrections. However, the majority of differences are small and only indirectly linked to corrections and therefore not easy to explain without a prohibitively large amount of analysis.

Is it possible for a player to have an official rating but no original rating?

Yes, as a result of human error in the subscription process. For example, suppose a player subscribes by post but the membership officer moves house and forgets to pick up his old mail (this has actually happened). Or suppose a player hands cash to an ICU committee member at a tournament who then forgets to pass it on to the membership officer (this has also happened). Unless the error is discovered before the end of the month of publication, the player won't appear in the rating list and therefore, when the error is eventually corrected, no original rating will exist. The recommended way to subscribe is online which minimises the potential for human error.