Re: [Bug-gnubg] Re: Naw..panic or teach me what I am doing wrong !!

[Neil Kaz]

Anyhow..I'm glad that this will be fixed and look forward to testing it. But
I did look at every play in 20 of the games you sent me from my basic 31
backgame. My rating of GNU's play would be 93 out of 100 vs my rating for SW
4 of 95 out of 100. SW4 makes some slight errors and my sincere opinion is
that GNU makes a couple more. All in all, I was extremely pleased with GNU's
play from here, and I honestly don't think any other bot would even rate an
80 from this position.

Once we get the fixed version we can again run the rollout and be certain
that GNU can get close to Snowie's results
(like when I did an 11 move truncated rollout)

After that, my next step is to take a look at GNU vs SW 4 evaluation of
holding game /quasi backgame type positions. SW 4 tends to overvalue the
defender's chances in those position and I am hoping GNU does a more
accurate job.

..neilkaz..
----- Original Message -----
From: "Joseph Heled" <address@hidden>
To: "gnubg" <address@hidden>
Cc: "'Neil Kaz'" <address@hidden>
Sent: Thursday, January 09, 2003 8:29 PM
Subject: Re: [Bug-gnubg] Re: Naw..panic or teach me what I am doing wrong !!



Fix has been submitted. I hope Nardy or Oystein can build a new windows
version soon so you can all test it. I think it will turn out that GNUbg
at 2ply plays this better than SN 4, but I can be wrong.

-Joseph

Joern Thyssen wrote:
> On Fri, Jan 10, 2003 at 10:44:02AM +1300, Joseph Heled wrote
>
>>but now it gives me strange results. The difference seems to be when
>>using (or not using) the truncation. Can you try those 2
>
> [snip]
>
>>I get wildly different results. Perhaps the evaluation somewhere else is
>>  wrong?
>
>
> Try inverting the equivalent test in the bearoff truncation evaluations
> -- it's just an unfounded guess!
>
> Jørn
>
>
> _______________________________________________
> Bug-gnubg mailing list
> address@hidden
> http://mail.gnu.org/mailman/listinfo/bug-gnubg
>
>