                       |
| ||||
| XGID=--B-AB-BA-------ab-cccDCc-:1:1:-1:00:0:0:3:0:10 | |||||
Blue played 3/2 (White on roll) | |||||
 |
| ||||
| XGID=--AAABAAA-------ab-cccDCc-:1:1:-1:00:0:0:3:0:10 | |||||
Blue played 7/6 (White on roll) | |||||
If you widen your screen, the two positions above will appear side by side.
In the left-hand (first) position, Blue tightly played 3/2 with his ace. In the right-hand (second) position, he loosely played 7/6.
In both positions, White leaves a single blot with 66, and a double blot with 61. What no respondent seems to have been aware of is that White will (by eval) volunteer a single blot with 12 additional numbers (54 53 52 43 42 32).
After all 15 of the above numbers, Blue has a better chance in the left-hand (first) position to gain control of the outfield and make some immediate progress against the blot he hits (wins more, gets gammoned less), than in the right-hand position where chances are he'll be hit back more times and/or more often in the ensuing scramble, and White will be bearing off before Blue can improve his board.
Moreover, in the right-hand position, against Blue's six-blot mess, White will volunteer two blots (pay now) with 8 ace-x numbers (51 41 31 21) in order to bear in more safely next roll, whereas in the left-hand position she will keep the 8pt (pay later). Double-edged -- true, but this represents additional net equity for White in the right-hand position.
No doubt XG misplays later hit-and-contain variations, transpositionally making the 2pt (or whatever) when it shouldn't, but it's hard to imagine it would negate the sizable rollout margin (.065) favoring Blue's solid 3/2 move. This is exactly the wrong time to play loosely, however strong the temptation might be to build the board from the highest point down.
Nack