Is bot analysis of manually altered positions reliable?

[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums

Is bot analysis of manually altered positions reliable?

Posted By: Karol Szczerek
Date: Saturday, 31 October 2015, at 8:58 p.m.

In Response To: Is bot analysis of manually altered positions reliable? (Timothy Chow)

The positions posted are just a quick illustration of the problem in question. I don't think bot's wrong here, since moving the blot 1 pip back is a obvious blow to our blitzing potential.
The position below is likely to emerge from legal play.

White is Player 2

score: 0
pip: 137

Unlimited Game
Jacoby Beaver

pip: 140
score: 0

Blue is Player 1

XGID=---bBBCBB---bB---bcbbb--B-:0:0:1:00:0:0:3:0:10
Blue on roll, cube action?

eXtreme Gammon Version: 2.19.208.pre-release

The variant below, created by moving bottom player's 8pt. to 9pt., isn't as likely to be reached in normal game.

White is Player 2

score: 0
pip: 137

Unlimited Game
Jacoby Beaver

pip: 142
score: 0

Blue is Player 1

XGID=---bBBCB-B--bB---bcbbb--B-:0:0:1:00:0:0:3:0:10
Blue on roll, cube action?

eXtreme Gammon Version: 2.19.208.pre-release

My point is, the bot's evaluation of strength of the broken prime here may be off, because it's an unusual thing to encounter in a midgame prime battle like this. And bot's evaluations are based on relations between more typical sets of positional factors (neurons).
Just like, on the extreme end, the super backgames and completely custom positions like the one below are most probably completely misjudged by bot.

White is Player 2

score: 0
pip: 210

Unlimited Game
Jacoby Beaver

pip: 165
score: 0

Blue is Player 1

XGID=c----EBb-cc--B-B-B-B---bb-:0:0:1:00:0:0:3:0:10
Blue on roll, cube action?

eXtreme Gammon Version: 2.19.208.pre-release