This position has an interesting story – I’ve played this match as part of my attempt to find out how strong the really strong bots (XG Mobile, True Backgammon/BGBlitz, Backgammon NJ) really are. I do this by feeding a good number of matches into the three top PC bots (XG2, BGBlitz, Gnu BG) and looking at their analysis at their deepest level.
On this one they totally disagree. Which also means that this is the one position where I won’t offer a solution to you 🙂
All agree on one thing: Blue has a strong lead, with 81% winning chances for blue but low 3% gammon chances. But this is a 2-away, 4-away situation. If blue doubles, white will immediately re-double and cube ownership will have no value for blue. Should blue still double?
- BGNJ didn’t double
- For XG2 (Roller++) this is a clear double and a close pass. Not doubling would be a 0.141 blunder. Taking would be a 0.014 minor error.
- For BGBlitz (5ply) this is almost no double, and a clear take. Not doubling would be a 0.006 mistake, ie. not a mistake at all. Passing would be a big 0.143 blunder.
- Gnu Backgammon (4ply) finds itself between these extremes: Double/Take, not doubling is a -0.049 error, passing is a 0.018 error.
XG2’s rollout interestingly made it less of a blunder to not double (-0.105), but a clearer pass – taking now would be a 0.034 error.
So, I leave you with these figures. Being a novice compared to the bots I won’t try to decide what’s right.
Oh, btw., one move later we arrive at this position:
… in which the bots disagree just as much: XG2 says it is a 0.188 big blunder not to double, while BGBlitz says this is borderline double/no double (-0.004 error not to double).
And one move later, BGNJ rolled 55, could bring its checkers from 13 to its prime, doubled and I passed (to which XG2 and BGBlitz finally agreed, and this time gnu disagreed, considering it too good to double 🙂 )