[Top][All Lists]

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

Re: [Bug-gnubg] matchplay? moneyplay? bug? or mistake?

From: Rod Roark
Subject: Re: [Bug-gnubg] matchplay? moneyplay? bug? or mistake?
Date: Thu, 4 Mar 2004 11:16:36 -0800
User-agent: KMail/1.6

I'm not sure if this applies, but note that there are cases
where a post-crawford double would be automatic for a human
player but merely "optional" for a bot.  I observed this
when experimenting with ParlorBot's interpretation of GnuBG

In particular, if there is no set of possible rolls between
now and your next turn where your opponent would refuse a
double at that time, then there is no reason to double now.

-- Rod

On Tuesday 02 March 2004 12:22 pm, don wrote:
> Hi,
> I'm new to this list so I may be asking something that has been addressed.
> I have been playing with an interface for the external function and fibs ..
> most everything seems to be working well.
> The bot seems to be playing well except that it doesn't ever automatically
> double when it is trailing post crawford.
> I am looking into the possibility that something is reversed in the board
> because I adjusted for the board reversal by hand, by simply reversing the
> moves before sending thm back to fibs... but I'm pretty sure that it is not
> looking at the wrong player because it seems it always has an advantage and
> market losers when it doubles. So another possibility is that it is playing
> a money game when it should be playing a match and not using equity tables
> etc. at all.  Could I simply be not setting it up right? I assumed that it
> should get the matchscore information from the fibsboard and double
> accordingly. Do I need to set it up for matchplay? Or am I maybe dealing
> with the reversed board issue in the wrong way.
> Don

reply via email to

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