[gnugo-devel] twogtp match

[Paul Pogonyshev]

i've run a twogtp match between 3.3.17-pre-7 (white) and 3.2 consisting
of 100 games with handicap 2 and komi 0.5 (i set it to 5.5, but the
script had different opinion).

black won 56 to 44. if the komi has been 5.5 as intended then (assuming
it wouldn't have influence the games which it doubtly would have much),
white would have won 58 to 42. hence, i think, the current cvs is
approximately 1.2 - 1.3 stones stronger than 3.2.

here is the distribution of results (each column represents 3 adjacent
results e.g 0.5, 1.5 and 2.5):

                                    | *
                                    | *
                                    |**
                                *   |**
                                ** *|** **
                               *****|********
                            * ******|********        *
                       * *  ********|*********   *   **
       *    *  * *   ****** ********|********* ********
<---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+--->
  W                                                                   B

and the results themselves:

B+16.5
W+2.5
W+0.5
W+86.5
W+10.5
W+6.5
W+1.5
B+26.5
B+49.5
B+4.5
B+12.5
W+7.5
B+10.5
W+13.5
B+11.5
B+9.5
W+22.5
B+10.5
B+16.5
B+1.5
B+45.5
B+0.5
W+17.5
W+4.5
W+19.5
W+5.5
B+20.5
W+3.5
B+3.5
B+7.5
B+2.5
B+2.5
B+18.5
B+2.5
W+7.5
B+13.5
W+22.5
B+24.5
W+11.5
W+14.5
W+35.5
W+8.5
B+15.5
W+9.5
B+1.5
B+3.5
W+11.5
B+18.5
W+17.5
B+5.5
B+8.5
B+9.5
W+10.5
W+70.5
W+4.5
B+3.5
W+39.5
B+43.5
W+38.5
B+21.5
B+31.5
B+34.5
B+13.5
W+32.5
B+41.5
B+6.5
B+51.5
W+0.5
W+60.5
B+1.5
B+4.5
B+38.5
W+37.5
W+16.5
B+21.5
B+14.5
B+4.5
B+51.5
W+21.5
B+36.5
W+0.5
B+3.5
W+56.5
B+7.5
W+30.5
B+50.5
B+16.5
W+43.5
W+12.5
W+9.5
W+12.5
B+5.5
B+20.5
B+22.5
B+48.5
W+20.5
W+8.5
B+22.5
B+13.5
W+27.5

Paul