gnugo-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[gnugo-devel] Recent games against Many Faces


From: Gunnar Farneback
Subject: [gnugo-devel] Recent games against Many Faces
Date: Tue, 19 Mar 2002 19:57:51 +0100
User-agent: EMH/1.14.1 SEMI/1.14.3 (Ushinoya) FLIM/1.14.2 (Yagi-Nishiguchi) APEL/10.3 Emacs/20.7 (sparc-sun-solaris2.7) (with unibyte mode)

Many Faces has been playing on NNGS for a number of days now and I've
set up two games against GNU Go. Both can be found on the recent games
page. (The second game is currently adjourned at the finished
position. Hopefully I can get it through the scoring soon.)

The first game was even with GNU Go playing black. GNU Go made some
weak moves in the opening but somehow managed to get a lead in the
middle game. However, it botched the lead by first allowing some
stones in the center to be cut off and die. Later the lower right
corner was also cut off and after GNU Go had misread the life and
death it got killed. Many Faces won by 38.5.

In the second game GNU Go played black with two stones handicap. This
game featured very little fighting and a lot of slow moves from both
programs, but mostly from GNU Go. Close to the end GNU Go was leading
by a tiny margin but Many Faces blundered and GNU Go could capture 14
stones in an atari-atari sequence to get a 30.5 point win.

Some random observations:
1. Many mistakes can be attributed to incorrect connection analysis.
   There is great potential for improvement when we integrate
   readconnect with the dragon amalgamation (post 3.2).
2. The last time GNU Go played against Many Faces, MF was allowed to
   make many points in the center by building a moyo undisturbed.
   Arend's influence revisions seem to have caused substantial
   improvements in this area of the game.
3. The fuseki and early middle game are most in need of tuning.

/Gunnar



reply via email to

[Prev in Thread] Current Thread [Next in Thread]