play online chess
| What alberlie means... |
...is that this question has been canvassed before. The fact is that though the statstical system (ELO) is the same, and the numerical values measure relative (note that: 'relative') chess playing strengths, the two are not really comparable. They are actually measuring different things.
The FIDE system orders chess players who participate in international OTB (over-the-board) games under strict tournament rules.
GK measures those who play chess by internet on this particular site. The very fact that the systems were instituted at different times - roughly 30 years' difference - is alone sufficient to create a significant difference in the numbers you are looking at. I can think of all sorts of other reasons for such differences. Bear in mind, as I've said elsewhere, the rating number has no meaning in itself. I seriously doubt Professor Elo himself would say anything different. It is where your rating stands on the list that is significant, and the size of the difference between your rating and another's (even that should be taken 'cum grano salo'!). Of course, the whole thing is built on the ongoing history of all the game outcomes (W, D, L, and who played whom) since the rating system in question was instituted (FIDE in the 1960s, GK c.2000).
Personally, I have my doubts about the ELO system, about the claims made for it, the uses to which it is put, and the way it is interpreted. I'm pretty certain it isn't really robust enough mathematically to update one's rating game by game as it now does (the original FIDE system didn't). But as also has been said elsewhere, it's probably the best we can hope for.