When I first discovered there was such as thing as backgammon bots (bought one of the versions of JellyFish), I was intrigued by the bot's propensity to hit twice in its inner board whenever it could.
Playing the the GNUbg-based bots on FIBS has taught me the value of such double hits. However, it looks like one can go too far in applying the principle. See below.
I saw the opponent had made no points in his inner board, I had a 3-point board, so I figured it was time to hit twice. Wrong, according to the bot.
When I later looked at the position, I noted that my opponent's 3s and 1s are duplicated--a bad thing for him. Wondering if that's why it makes sense to hit and cover the 5 point and hope to bang the checker on the 9 point next turn.
The score (after 1 game) is: diegobello 2, You 0 (match to 3 points, Crawford game)
Move number 9: You to play 31
| diegobello | 166 |
|---|---|
| |
| You | 139 |
Position ID: 4HPwCCFmTuIJIA Match ID: 8IllACAAAAAA | |
• You moves 10/9* 8/5*
Alert: very bad move ( -0.108)
| # | Ply | Move | Equity | |
|---|---|---|---|---|
| 1 | 4 | 8/5* 6/5 | +0.611 | |
| 76.2 41.1 3.3 - 23.8 4.5 0.1 | ||||
| 4-ply cubeful prune | ||||
| • | 2 | 4 | 10/9* 8/5* | +0.503 ( -0.108) |
| 71.1 36.9 3.1 - 28.9 6.5 0.3 | ||||
| 4-ply cubeful prune | ||||
| 3 | 0 | 8/5*/4 | +0.380 ( -0.231) | |
| 66.1 31.7 2.0 - 33.9 7.7 0.2 | ||||
| 0-ply cubeful prune [expert] | ||||
| 4 | 0 | 13/10 6/5* | +0.364 ( -0.248) | |
| 65.0 30.5 2.4 - 35.0 9.1 0.4 | ||||
| 0-ply cubeful prune [expert] | ||||
| 5 | 0 | 13/9* | +0.360 ( -0.251) | |
| 64.9 29.6 2.3 - 35.1 8.5 0.3 | ||||
| 0-ply cubeful prune [expert] | ||||
| 6 | 0 | 24/21 10/9* | +0.351 ( -0.260) | |
| 64.9 26.8 2.0 - 35.1 8.0 0.3 | ||||
| 0-ply cubeful prune [expert] | ||||