On February 28, 2023 at 7:09:53 AM UTC-7, Timothy Chow wrote:
On 2/27/2023 5:06 PM, Robert Zimmerman wrote:
XGID=-ABbBCC-B----a---bbbbBbb--:0:0:1:54:0:0:3:0:10
X to play 54
I'd break anchor here ... And we gain a lot if we are not hit.
I'm surprised that more people haven't participated in
this thread which I thought would be very interesting.
It looks like all the bozos of RGB have they gone Walt
on me but I won't retaliate... ;)
As I have pointed out countless times over the years
that by their nature, even the gamblegammon playing
bots don't gamble.
It may be interesting for humans to discuss "pay now
or later", "safe or bold", etc. it's totally idiotic to consult
bot rollouts because of the plain fact that bots always
pay later, always play safe, etc.
The correct play here is 6/1 5/1 and it's also the play
referred to as the "safe" one here. Thus, any/all other
moves can then be said to be the "bold" one/s.
I'm not necessarily saying that Tim or other humans
who wouldn't play 6/1 5/1 here are all wrong. In fact,
it may very well be the opposite and that the bot may
be wrong because of its inability adjust its play to the
score (let alone to the skill level of its opponent).
Here are some examples to show how the bot/s will
play 6/1 5/1 regardless of the cube value in a money
game. (I will post separately on match games.)
All hints from analysis level set to XGR++.
Position in initial post with jacoby on:
6/1 5/1 eq:-0.0472 [49.1% 8.5% 0.1% - 50.9% 9.9% 0.3%]
21/16 5/1 eq:-0.2177 (-0.1705) [43.7% 12.0% 0.5% - 56.3% 15.0% 0.4%]
With jacoby off:
6/1 5/1 eq:-0.0488 [49.1% 8.6% 0.1% - 50.9% 9.9% 0.3%]
21/16 5/1 eq:-0.2174 (-0.1686) [43.8% 10.6% 0.2% - 56.2% 14.6% 0.4%]
I don't know why "game winning chances" are different
even if by very little but jacoby doesn't make difference
enough in this position to not play safe. Once the cube
is turned, all "game winning chances" remain the same
regardless of the cube value.
With X owning cube at 2:
6/1 5/1 eq:+0.2766 [49.1% 8.6% 0.1% - 50.9% 9.9% 0.3%]
21/16 5/1 eq:-0.0388 (-0.3153) [43.8% 10.6% 0.2% - 56.2% 14.6% 0.4%]
With X owning cube at 32:
6/1 5/1 eq:+4.4254 [49.1% 8.6% 0.1% - 50.9% 9.9% 0.3%]
21/16 5/1 eq:-0.6201 (-5.0456) [43.8% 10.6% 0.2% - 56.2% 14.6% 0.4%]
With O owning cube at 2:
6/1 5/1 eq:-0.4250 [49.1% 8.6% 0.1% - 50.9% 9.9% 0.3%]
21/16 5/1 eq:-0.6973 (-0.2723) [43.8% 10.6% 0.2% - 56.2% 14.6% 0.4%]
With O owning cube at 32:
6/1 5/1 eq:-6.7996 [49.1% 8.6% 0.1% - 50.9% 9.9% 0.3%]
21/16 5/1 eq:-11.1564 (-4.3568) [43.8% 10.6% 0.2% - 56.2% 14.6% 0.4%]
Setting cube limits don't make any difference either.
Playing a 1-point match, i.e. without risking to lose a
gammon or backgammon, doesn't make a difference.
1-away/2-away/crawford combinations don't matter.
Do you think bots are capable of making "bold" plays
depending on score and/or cube value in matches..?
We will see in my next post...
MK
--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)