From: Larry Kaufman <lkaufman <at> WIZARD <dot> NET>
Date: 11 jan 2000
Subject: Re: Moscow Shogi OPEN 2000 and Kyu-Dan Test

>>      In case you didn't notice, two of the Japanese players who played in
>> Russia also played in London last summer, and so have ratings.  So there is
>> some (small) basis for rating the Russian event.
>>      As to the "kyu test", read the below letter from Mr. Nosovsky:
>> Kyu-Dan Test
>> First, Tatsutomi, Muro and Moro played equal games against each player
>> (sometimes two players).
>> Those who get ranks 10,11,12, 13, 14, 15 kyu played only Mr. Tatsutomi,
Muro
>> and Moro.
>> After that they recomended some player (who  get better then 10 kyu rank to
>> play with Mr. Nakao (6 Dan Pro).
>> Mr. Nakao have played with 3 players semulatively and gaves kyu from 9-8 .
>> All Mr. Tatsutomi, Moro and Muro have played games against Nikonov and say
>> that he is rather strong.
>> Two players Mr. Nikonov and Mr. Shpilev were selected to play with Mr.
Nakao
>> equal game one by one , and then he say that they have to play again.
>> Nikonov against Shpilev and Mr. Nakao sitting near and write their game.
>> He will show this game to Japanese Shogi Federation.
>> After Nikonov won , and Mr. Nakao decided that he is above 1 kyu-1 Dan ,
>> then 4 Japanese players discuss it and decided about 1 Dan.
>> Mr. Kisluk was recognized as 5 kyu during 1-st International Tournament in
>> Japan (you probably have met with him).
>> Japanese played with him and upper him 1 kyu.
>> Regards Nosovsky
>>
>>      Finally, back to our Pan-Atlantic list, I wonder if you might try a
>> run using floating-point numbers.  It shouldn't matter much if everything
>> is rounded, but if fractions are being truncated it could cause a serious
>> bias.  I think something must be wrong, because the value of bishop in the
>> free-floating run seems lower to me than makes sense from the data; perhaps
>> this is the problem.  I know that with floating point numbers the ratings
>> on the first run will never completely converge, because some players lost
>> every game, but this won't matter as those players won't influence the
>> handicap values, which is all that matters on the first run.
>>
>>      Larry
>
     Sorry, this was supposed to be a private email, I didn't mean to send
it to the List.

     Larry