Arpad Elo’s rating system initially sought to compare players across eras. Instead it has become the keystone of chess organisation. Titles, invitations and appearance fees are all benchmarked to rating.
Ironically, ratings aren’t good at normalising historical performance. Over 39 years, there has been inflation as the pool of highly-rated players has grown. Other practical problems are apparent. Ratings are frozen between lists, which means for minimum periods of 3 months.
Changes are calculated through assignment of weights, called “K factors”. Ratings don’t change quickly to reflect shifts in form, or the meteoric rise of prodigies like Carlsen. This has practical importance since invites (even to world title qualifiers) and fees are rating-driven.
Fide plans to experiment with an increase of the K factor. It will run parallel lists with the current official K and a new higher K until July 2011 when there will be a review. This has sparked fierce debate among the mathematically-competent. A higher K will cause sharper rating shifts. While this reflects current form better, it could lead to manipulation.
The diagram, WHITE TO PLAY, (Aronyan Vs Leko, Nalchik Grand Prix 2009) is a key moment in the last round clash that decided first place. Leko had drummed up what seems a menacing attack with white forced to weaken his pawns.
Now Aronyan played 34.Re5! Nxe5 35.Rxe5 f5 36.Bb3 Nd5 37.Rxe6 Kh8. The exchange sacrifice leaves white with two bishops and an extra pawn. The c-d passers dominate and black’s queen is offside. The game is essentially over.
The mop up continued 38.Qe1 Nf6 39.Qe5 Re8 40.c6 Rbc8 41.Qxb5 Qg6 42.h5 Qxh5 43.Bf4 a6 44.Qxa6 Nh7 45.c7 Ng5 46.Rxe8+ Qxe8 47.d5 Ra8 48.Qc4 Kh7 49.d6 Qe1 50.Qf1 Qe8 51.Qd3 Qd7 52.Qc4 Qe8 53.Bxg5 hxg5 54.Qg8+! (1-0). The finish is elegant but there were many less spectacular routes to victory.