RE: [Bug-gnubg] Optional redouble, or is it automatic

[Ian Shaw]

> -----Original Message-----
> From: Joern Thyssen [mailto:address@hidden
> Sent: 09 September 2002 13:28
> To: Jim Segrave
> Cc: Ian Shaw; GnuBg Bug (E-mail)
> Subject: Re: [Bug-gnubg] Optional redouble, or is it automatic
> 
> 
> On Mon, Sep 09, 2002 at 01:19:57PM +0200, Jim Segrave wrote
> > On Mon 09 Sep 2002 (11:10 +0100), Ian Shaw wrote:
> > > Why is this type of position called an "optional 
> redouble"? Surely the
> > > redouble is automatic. I guess the 2-ply analysis is 
> looking ahead and
> > > automatically doubling on the next roll so it sees no 
> loss of equity.
> > > 
> > >     GNU Backgammon  Position ID: 2W4AAGdsbRMBGA
> > >                     Match ID   : AgGgAAAAEAAA
> > 
> > Odd - using today's build with evaluation set to world-class++ for
> > chequer and cube play I get the following evaluation:
> > 
> > No redouble         : +1.201
> > Redouble, take      : +1.347   (+0.146)
> > Redouble, pass      : +1.000   (-0.201)
> > 
> > Correct cube action: Too good to redouble, pass
> > 
> > Which, to my untrained eye looks far more reasonable.
> 
> O is on roll... 15% looks reasonable with 2 chequers on the 
> bar. Did you
> reverse the position to get X on roll?
> 
Just to clarify the position:

    GNU Backgammon  Position ID: 2W4AAGdsbRMBGA
                    Match ID   : AgGgAAAAEAAA
    +-1--2--3--4--5--6-------7--8--9-10-11-12-+  O: mlp (Cube: 4)
    |    X  O  O  O  O |   | O  O     O       |  0 points
    |    X  O  O     O |   | O  O             |  On roll
    |    X             |   |                  |  
    |                  |   |                  |  
    |                  |   |                  |  
    |                  |BAR|                  |^ 5 point match
    |                  |   |                  |  
    |                  |   |                  |  
    |             X    |   |                  |  
    |    O  X  X  X  X | X |                  |  
    | X  O  X  X  X  X | X |                O |  2 points
    +24-23-22-21-20-19------18-17-16-15-14-13-+  X: ian

This position is ID is actually bG0TARjZbgAAZw on my display. The score and
cube positions are correct, but the chequer positions are reversed. I have
two chequers back behind a solid four-prime. My opponent has three on the
two point and two on the bar.

Uncharacteristically I am behind in my updates and this is Oystein's
20020814 build, so this may be fixed by now.

The correct position is this:
 
   +-1--2--3--4--5--6-------7--8--9-10-11-12-+   O: ian
    |    X  O  O  O  O |   | O  O     O       |  2 points
    |    X  O  O     O |   | O  O             |  
    |    X             |   |                  |  
    |                  |   |                  |  
    |                  |   |                  |  
    |                  |BAR|                  |^ 5 point match
    |                  |   |                  |  
    |                  |   |                  |  
    |             X    |   |                  |  
    |    O  X  X  X  X | X |                  |  On roll
    | X  O  X  X  X  X | X |                O |  0 points
    +24-23-22-21-20-19------18-17-16-15-14-13-+  X: mlp (Cube: 4)

Reconsidering my original reason for posting, Os takepoint is at 75% (W-H
MET) because he can drop to Crawford 3-away. Nis is correct to argue that X
might drop incorrectly in future so holding on to the cube is OK. I'm not
sure that a bot sees this though since it assumes correct play by both
sides. In that case the only drops you get are correct drops because O has
lost his market, and therefore O wishes he had doubled earlier.

Ian