gnugo-devel
[Top][All Lists]
Advanced

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

Re: [gnugo-devel] Bad game


From: Paul Pogonyshev
Subject: Re: [gnugo-devel] Bad game
Date: Sat, 11 Dec 2004 19:00:35 +0200
User-agent: KMail/1.4.3

Dan wrote:
> > > > GNU Go (3.7.1 from CVS) just played an exceptionally bad game against
> > > > me. 9x9 game, 2 stones handicap. I will analyze it further later this
> > > > weekend, but I wanted to show it anyway.
> > > >
> > > > The comments here are my own, so take with a grain of salt, I am only
> > > > 5kyu.
> > >
> > > Is this Schokobot?
> >
> > I believe Heikki played with a local GNU Go, because his SGF file was
> > created with Quarry.  You cannot connect to any servers with Quarry
> > (yet), only play locally via GTP.
>
> Sorry, I somehow got the idea this game was played on KGS ...
> obviously it wasn't.
>
> I don't understand how B resigned since the boardsize is not 19x19.

Firstly I thought Quarry somehow managed to misinterpret GNU Go's
responses, but:

address@hidden quarry]$ ~/gnugo/interface/gnugo --mode gtp -l ~/heikki.sgf
genmove black
= resign

Seems like `resigning.html' should be revised.  If this behaviour seems
undesirable, we should also revise GNU Go itself.

In `genmove.c', function should_resign():

  /* We resign 19x19 games only, smaller board games are fast enough.
   * We resign only if the margin is bigger than 45 pts and if we are
   * behind (of course).
   *
   * As an exception to this rule, we resign on any board size if
   * it looks like all our dragons are dead and the generated move
   * is a pass.
   */
  if (move == PASS_MOVE && !lively_dragon_exists(color))
    return 1;

Paul





reply via email to

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