I'm beginning to get a feel for this position, but it still remains somewhat mysterious to me. If you think 24/20 8/5 is "obvious" in retrospect, then test your intuition against the following rather similar position.
| team | 162 |
|---|---|
| |
| box | 160 |
Position ID: 4HPKATDg88gBMA Match ID: cIkRAAAAAAAA | |
Here, if gnubg is to be believed, 24/20 8/5 is a lot worse (-0.215) than 13/9 8/5 (-0.133) or 24/21 13/9 (-0.134). I guess safety is still important, and the duplication of 4's in the original position is critical.
By the way, I'm still interested in what other bots think, because gnubg and bgblitz don't entirely agree, e.g., gnubg thinks 24/20 11/8 is much worse (by -0.074) than 24/20 8/5 in the original position, whereas bgblitz thinks 24/20 11/8 is the third-best move.
As for Marty Storer's question, gnubg rates 24/21 24/20 at -0.221 in the original position and -0.235 in the new position, and 11/7 8/5 at -0.246 in the original position and -0.208 in the new position. In other words, neither play comes close to the top. During the game, I was on the team, but the box thought a long time, and in the meantime I studied 24/21 24/20. I think in practice this move will fare better than the bots might lead you to think, because there is a good chance that an immediate double-hitting decision will arise which anyone but an expert has a good chance of misplaying badly.