RE: [Bug-gnubg] Permit GNU to consider deliberate cube errors

[Ian Shaw]

I think Christian gets to the heart of the problem. It's not only
doubles that are an issue; it's the moves as well. It makes no sense to
implement one without the other.

Suppose an opening goes 43: 13/9 13/10 is a 0.02 error. Many experts
play this because it leads to more complicated games, where they feel
they have an advantage.

How can gnubg know if :
A) The player thought it was the best move.
B) The player played a non-best move because he wants to induce errors
from the opponent.
C) The opponent made a mistake induced by the player's error.
D) The opponent would have made that mistake anyway.

I see no way to read intent, so I think gnubg is best off scoring
players against "perfect" play (assuming perfect analysis, of course).

-- Ian

> -----Original Message-----
> From: address@hidden 
> [mailto:address@hidden 
> On Behalf Of Christian Anthon
> Sent: 25 July 2006 15:27
> To: Albert Silver
> Cc: address@hidden
> Subject: Re: [Bug-gnubg] Permit GNU to consider deliberate cube errors
> 
> Well I suppose you would like to implement bad non-doubles, 
> bad takes and bad moves as well. If you can come up with a 
> good way to measure these, I'm sure the idea stands a much 
> bigger chance of being put in to gnubg.
> 
> Christian.
> 
> On 7/25/06, Albert Silver <address@hidden> wrote:
> > I wasn't sure how to put this in the subject, but the idea 
> is simple enough:
> >
> > Allow GNU to consider deliberate cubing errors that are played in 
> > oredr to entice larger decisions from the opponent. What this would 
> > mean is that while it could give the absolute values of the equity 
> > lost when I doubled
> > (ex: doubling is a 0.150 blunder), and the equity lost when my 
> > opponent made a mistake in dropping (doubling might be a 0.150 
> > blunder, but my opponent makes an evenor taking, it would 
> also give an 
> > alternative grade/evaluation allowing for deliberate errors.
> >