Sorry to drag this subject up again. However, as it is one of the neatest things on Brainking, and is important to many people, I will excuse myself. On the Espionage board, we are currently having a debate on BKR and have posted the latest entry below.

BKR ratings are important. Having a tight BKR range at the top promotes competiveness and moral. However, where someone has leapt in with an extremely high BKR rating, causes abuse (defend rating by only playing low rank players) and creates less competiveness as that rating is almost unassailable as well as frustration in the other players.

The feature request is to modify the initial rating awarded during the first 4 games. There are many possible ways of doing this, but perhaps the easiest would be to create a cap on the number of points awarded for each game. For example, assuming all 4 games won, a cap of 100 could produce an initial provisional BKR of 1700, or a cap of 150 could produce 1900.

Latest post on thread on Espionage board:
The flaw is more than the abuse.
The ratings work very well after the initial provisional BKR has been awarded, it is how the initial rating is derived that is often specious.
There is a big difference between inaccurate and wildly spurious.
Here is a real example from earlier this year. You can find many similar examples across all games.

Game 1 won, opponent BKR between 1500 – 1600
Game 2 won, opponent BKR between 1400 – 1500
Game 3 won, opponent BKR between 1700 – 1800
Game 4 won, opponent BKR between 1500 – 1600

So what initial provisional BKR should be awarded?
1600 – 1700 Largely inaccurate, as we have information on a 1700-1800 win.
1700 – 1800 Inaccurate, but best we can do with information available.
1800 - 1900 Largely inaccurate, no information to validate this.
1900 – 2000 Grossly inaccurate, absolutely nothing to suggest this.
2000 – 2100 Wildly Spurious!
What was awarded? A figure half way between 2000 – 2100.

You decide, flawed or not?