[Computer-go] Could a 'doubling dice'** encourage early resignation by programs?

[Stefan Kaitschick]

Am 28.01.2011 12:15, schrieb Nick Wedd:
>> That sounds crazy, because it has such a gambling feel to it, but its 
>> true.
>> Its because even a mediocre player can muddle his way through the 
>> roll of the dice, but only a strong player can judge well what his 
>> current chances are.
>> The basic idea is that you need a 25% chance to win to be able to 
>> accept the double.
>> Because at 25% the expected value of accepting((2 - 3*2)/4) is the 
>> same as of declining(-1).
>
> This calculation is mistaken (you are forgetting that if he accepts 
> the cube then he will have it and you won't).  The correct value, 
> assuming that each move makes only a small difference to the winning 
> chances, is 80%.
Nice. Because you have a 1 in 4 chance to go from 20% to 80% yourself. 
And 80% has the same utility as 100%.
  So the "real" value is somewhere between 20% and 25% because your 
chances might jump past the 80% mark, lowering the utility of the 
doubling utensil.
Back to the game of go: instead of the backgammon mechanism, you could 
allow each side to double with every move, but for a progressive komi 
penalty of 10,20,40 points ect.
Something like that might work. The question remains, would it would 
mean to win 8x :-)
It seems to make sense only in the context of gambling.

Stefan