t-regress: Add 2011-06-09-jinen-pachi30s-2.sgf
[pachi.git] / t-regress / games / 2010-12-05-pachi-CzechBot.sgf
blob40707665b1fce3c460324d1140badbc3b28ef25f
1 (;GM[1]FF[4]CA[UTF-8]AP[CGoban:3]ST[2]
2 RU[Chinese]SZ[9]KM[7.50]TM[540]OT[25/60 Canadian]
3 PW[pachi]PB[CzechBot]BR[1k]DT[2010-12-05]GC[34 +E1:bad! pachi does not see snapback after black B2+C1. The "does not seen snapback" bug is nasty.  The problem is that there is also a "global capture heuristic" if there're no fun local moves, it looks for groups to capture and if white tenukis and does not capture immediately, this heuristic kicks in.  But there is no "global atari heuristic" that would drive black similarly to play at F1 and create the snapback if I test with playout=moggy:capturerate=0, it reliably destroys the snapback i.e. with capturerate=90, white is much more likely to come back to the situation later even when playing tenuki first.]PC[The KGS Go Server at http://www.gokgs.com/]C[pachi [-\]: GTP Engine for pachi (white): Pachi Distributed Engine version 6.99 (Chihaku-devel): I'm playing the distributed engine. When I'm losing, I will resign, if I think I win, I play until you pass. Anyone can send me 'winrate' in private chat to get my assessment of the position.
4 CzechBot [1k\]: GTP Engine for CzechBot (black): MoGo version MoGo version 4.86.Soissons: I'll resign if you win and playing until you pass if I win. If you passed and I did not, it often means that the status of some stones are not clear enough for me (I want to be sure we will agree on dead stones). Clarify them. Maybe I am losing, but have not enough time to be sure I should resign. In this case play until the end :'( (very rare cases)
5 ]RE[B+Resign]
6 ;B[ee]BL[539.73]
7 ;W[ec]WL[539.38]
8 ;B[fc]BL[539.481]
9 ;W[fd]WL[538.761]
10 ;B[ed]BL[539.25]
11 ;W[gc]WL[522.602]
12 ;B[fb]BL[538.589]
13 ;W[dc]WL[506.866]
14 ;B[cd]BL[538.286]
15 ;W[gd]WL[468.604]
16 ;B[gb]BL[484.795]
17 ;W[gf]WL[454.735]
18 ;B[cc]BL[432.429]
19 ;W[hb]WL[415.258]
20 ;B[fg]BL[384.376]
21 ;W[ff]WL[413.67]
22 ;B[ef]BL[342.41]
23 ;W[cb]WL[412.103]
24 ;B[ea]BL[304.838]
25 ;W[db]WL[400.073]
26 ;B[bb]BL[270.898]
27 ;W[bc]WL[370.904]
28 ;B[ba]BL[242.592]
29 ;W[eb]WL[369.315]
30 ;B[gg]BL[216.047]
31 ;W[dg]WL[345.144]
32 ;B[eg]BL[193.483]
33 ;W[bf]WL[335.548]
34 ;B[cf]BL[173.592]
35 ;W[cg]WL[326.317]
36 ;B[be]BL[154.774]
37 ;W[eh]WL[317.596]
38 ;B[fh]BL[138.014]
39 ;W[ei]WL[316.158]
40 ;B[bh]BL[123.957]
41 ;W[bg]WL[307.42]
42 ;B[ci]BL[110.923]
43 ;W[ch]WL[298.909]
44 ;B[dh]BL[99.399]
45 ;W[fa]WL[298.242]
46 ;B[fi]BL[88.9]
47 ;W[bi]WL[270.732]
48 ;B[ai]BL[79.818]C[jlg [?\]: score went down from 86 to 35% :-(
50 ;W[di]WL[246.388]C[jlg [?\]: F9 big blunder
52 ;B[dh]BL[71.924]
53 ;W[af]WL[225.561]
54 ;B[bi]BL[65.008]
55 ;W[dd]WL[206.188]
56 ;B[bd]BL[59.124]
57 ;W[ac]WL[199.597]
58 ;B[de]BL[53.314]
59 ;W[ae]WL[181.457]
60 ;B[ce]BL[48.707]
61 ;W[ad]WL[167.546]
62 ;B[ga]BL[44.707]
63 ;W[ha]WL[162.808]
64 ;B[df]BL[40.68]
65 ;W[ab]WL[153.401]
66 ;B[ag]BL[34.665]
67 ;W[ah]WL[147.124]
68 ;B[di]BL[29.121]
69 ;W[fa]WL[145.747]
70 ;B[ag]BL[24.296]
71 ;W[hh]WL[136.612]
72 ;B[aa]BL[20.085]
73 ;W[hg]WL[135.716]
74 ;B[fe]BL[16.81]
75 ;W[ge]WL[135.117]
76 ;B[hi]BL[15.724]
77 ;W[ii]WL[125.385]
78 ;B[ih]BL[15.074]
79 ;W[ca]WL[119.001]
80 ;B[ii]BL[14.525]
81 ;W[gi]WL[116.748]
82 ;B[gh]BL[14.158]
83 ;W[hf]WL[114.599]
84 ;B[gi]BL[13.01]
85 ;W[ig]WL[112.506]
86 ;B[fb]BL[12.538]
87 ;W[fc]WL[110.445]
88 ;B[ga]BL[11.845]
89 ;W[id]WL[108.475]
90 ;B[bc]BL[11.013]
91 ;W[ia]WL[101.386]
92 ;B[fa]BL[9.713]
93 ;W[da]WL[99.696]
94 ;B[ic]BL[8.766]
95 ;W[hc]WL[96.15]
96 ;B[ie]BL[7.782]
97 ;W[ib]WL[94.699]
98 ;B[he]BL[7.363]
99 ;W[gb]WL[93.284]
100 ;B[hd]BL[6.471]
101 ;W[fa]WL[91.908]
102 ;B[ic]BL[5.902]
103 ;W[dg]WL[86.55]
104 ;B[cg]BL[5.091]
105 ;W[id]WL[82.556]
106 ;B[bf]BL[4.266]
107 ;W[if]WL[78.947]
108 ;B[ab]BL[3.591]
109 ;W[ei]WL[72.841]
110 ;B[eh]BL[2.715]
111 ;W[he]WL[67.609]
112 ;B[]BL[2.714]C[chid0ri [?\]: why didnt pachi resign earlier?
113 chid0ri [?\]: i wonder why did it play that blunder
114 chid0ri [?\]: ^ pasky
115 jlg [?\]: there are two bugs: one which caused F9, another one is that it should have resigned earlier
116 chid0ri [?\]: do you have logs?
117 jlg [?\]: yes I will send you logs for this one
118 chid0ri [?\]: thx
119 jlg [?\]: but I don't keep everything, it's too verbose with 320 cores
120 chid0ri [?\]: :)
121 jlg [?\]: you might be able to reproduce the bug even on a single machine
122 jlg [?\]: it just didn't see the snapback
123 chid0ri [?\]: well i'm more curious about why didnt it resign right now
124 chid0ri [?\]: ah
125 jlg [?\]: the snapback issue seems more serious to me than not resigning...