[Top][All Lists]

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

Re: [Bug-gnubg] Largest swing yet between 0-ply & 2-ply

From: Joseph Heled
Subject: Re: [Bug-gnubg] Largest swing yet between 0-ply & 2-ply
Date: Fri, 11 Apr 2003 07:28:45 +1200
User-agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.2) Gecko/20021202

So far, for any given net, it was always possible to find positions with large discrepancies. (large - greater than 2)


Ian Shaw wrote:
-----Original Message-----
From: Joern Thyssen
Sent: 08 April 2003 13:00

On Tue, Apr 08, 2003 at 12:45:56PM +0100, Ian Shaw wrote

This is the first time I've seen a swing of over 0.5 between 0-ply &
2-ply. I thought you, particularly Joseph, would be interested. It's
even larger on a centred cube, at 0.547.

If you look the at the cubeless equities they hardly change for bar/21
6/5: 0.231 to 0.217. The difference is much larger for the the safe play 0.063 to 0.347.

I don't find it particular disturbing. I think you can find many
positions where is really vital to look 2 or more plies ahead to get
accurate equities.

I don't find it particularly disturbing either. Your point about the safe play being almost unchanged is a good one. I posted it mainly because I know one of Joseph's training methods is to use 2-ply evaluations to train 0-ply. I think the slotted position is still an interesting example, even if the discrepancy reduces to "only" 0.284 cubeless. Particularly since it was a mistake made by a human in OTB play.
Just out of curiosity, what's the largest discrepancy you've found between 
0-ply and 2-ply? Joseph? Anyone?


Bug-gnubg mailing list

reply via email to

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