The "simple" answer is that overall white's rolls play worse when you eschew the hit and make the 3 point. White gains on any 5 or 42 but loses more on all other rolls.
I don't think any human could spot this play as the hit seems so intuitive. If you did consider it you would still have to run through white's responses and try to evaluate if it was indeed better.
          
 
          
 | | White is Player 2
score: 0
pip: 133 | Unlimited Game
Jacoby Beaver | pip: 136
score: 0
Blue is Player 1 |
|
| XGID=--a-CCCbB---A--b-b-ccCaa--:0:0:1:32:0:0:3:0:10 |
| Blue to play 32 |
| 1. | Rollout1 | 6/3 5/3 | eq: +0.208 |
| Player:
Opponent: | 52.90% (G:20.54% B:0.73%)
47.10% (G:9.55% B:0.33%) | Conf.: ±0.011 (+0.196...+0.219) - [100.0%]
Duration: 6 minutes 16 seconds |
|
| 2. | Rollout1 | 5/2* 4/2 | eq: +0.114 (-0.094) |
| Player:
Opponent: | 50.74% (G:19.52% B:0.54%)
49.26% (G:9.84% B:0.37%) | Conf.: ±0.011 (+0.102...+0.125) - [0.0%]
Duration: 7 minutes 10 seconds |
|
| 3. | Rollout1 | 21/18 4/2* | eq: +0.067 (-0.140) |
| Player:
Opponent: | 50.63% (G:19.43% B:0.71%)
49.37% (G:14.72% B:0.65%) | Conf.: ±0.013 (+0.055...+0.080) - [0.0%]
Duration: 7 minutes 30 seconds |
|
| 4. | Rollout1 | 21/16 | eq: +0.049 (-0.159) |
| Player:
Opponent: | 49.68% (G:16.39% B:0.60%)
50.32% (G:9.79% B:0.35%) | Conf.: ±0.010 (+0.039...+0.059) - [0.0%]
Duration: 7 minutes 45 seconds |
|
| |
1 1296 Games rolled with Variance Reduction.
Moves and cube decisions: 4-ply
Search interval: Large
|
eXtreme Gammon Version: 2.10