Game 1, move 12: Herd to move 6-1

[diane]

r_monk danced with 6-4, so here we are again...pick your fave move - or add anything more appealing..

3DYHA0Cbc6qAAA:UQmnAAAAGAAA

[ah_clem]

Spoiler

Making the 5 is better strategically, and by my count leaves one less return shot than making the 3.  Of course, I'd probably still make the five even if it left an extra return shot or two, but I'd have to think about it.  Here it's clear. 

[close]

[diane]

Another confusing one, where I almost nexted this - and then found making the 5 point is NOT coming up as the best option - any thoughts on why?

[stiefnu]

Spoiler

Another confusing one, where I almost nexted this - and then found making the 5 point is NOT coming up as the best option - any thoughts on why?

[close]

Why indeed - maybe Gnu was having a hissy fit?  According to Bgblitz on 3-ply, making the 5 point is by far and away the strongest play, with 9/8, 10/4 as the next best, at -0.043.

[ah_clem]

gnu 0 ply rollout

Spoiler

0 ply rollout prefers to make the 3 point.  I have no idea why.  I'm inclined to think that gnu is somehow mishandling this position. Steifnu says that bgblitz prefers making the 5 point, but he also say that it thinks 10/4 9/8 is better than making the three point, and I find that difficult to believe so I'm skeptical of bgblitz's result.

I'll try a 2 ply rollout.  It would be interesting to see results from other bots as well.



    1. Rollout          9/3 4/3                      Eq.:  +0.421
       0.763 0.428 0.001 - 0.237 0.071 0.004 CL  +0.421 CF  +0.421
      [0.001 0.009 0.000 - 0.001 0.001 0.000 CL   0.004 CF   0.004]

    2. Rollout          11/5 6/5                     Eq.:  +0.366 ( -0.054)
       0.746 0.393 0.003 - 0.254 0.085 0.006 CL  +0.366 CF  +0.366
      [0.002 0.009 0.001 - 0.002 0.001 0.001 CL   0.004 CF   0.004]

    3. Rollout          10/4 9/8                     Eq.:  +0.064 ( -0.357)
       0.596 0.223 0.003 - 0.404 0.088 0.004 CL  +0.064 CF  +0.064
      [0.002 0.007 0.001 - 0.002 0.001 0.001 CL   0.004 CF   0.004]
        Full cubeful rollout with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 900864643 and quasi-random dice
        Play: 0-ply cubeful prune [expert]
        Cube: 0-ply cubeful prune [expert]

[close]

[diane]

Why indeed - maybe Gnu was having a hissy fit? 

I think it was having a moment of clairvoyance..it knew what it was going to roll next...

[stiefnu]

0 ply rollout prefers to make the 3 point.  I have no idea why.  I'm inclined to think that gnu is somehow mishandling this position. Steifnu says that bgblitz prefers making the 5 point, but he also say that it thinks 10/4 9/8 is better than making the three point, and I find that difficult to believe so I'm skeptical of bgblitz's result.

Ooops, my apologies - no wonder it was hard to believe, I had a chequer misplaced.   

Bgblitz 3-ply analysis is now much the same as the gnu 2-ply rollout and likewise favours making the 3 rather than the more obvious 5 point:
1.   1. 0.876 mwp /  0.387           4-3, 9-3
2.   0.870 mwp /  0.354 (-0.007)  6-5, 11-5
3.   0.819 mwp /  0.102 (-0.058)  9-8, 10-4

[ah_clem]

Spoiler

Here's why making the three point is better:  

If we make the five point:

 GNU Backgammon  Position ID: m9tUgADcNgcDQA
                Match ID   : EQGrAAAAGAAA
+12-11-10--9--8--7-------6--5--4--3--2--1-+     O: r_monk
|             O  X | O | O  O  O  O       |     0 points
|             O    |   | O  O  O  O       |    
|             O    |   |          O       |    
|                  |   |                  |    
|                  |   |                  |    
^|                  |BAR|                  |     5 point match
|                  |   |                  |    
|                  |   |                  |    
|                  |   |       X          |    
| O                |   | X  X  X     X  X |    
| O     X  X  X    |   | X  X  X     X  X |     3 points
+13-14-15-16-17-18------19-20-21-22-23-24-+     X: You (Cube: 2)

I count 5 rolls that cover the three and make the closeout.


But if we make the three point:

GNU Backgammon  Position ID: 2+akgADcNgcDQA
                Match ID   : EYG5AAAAGAAA
+12-11-10--9--8--7-------6--5--4--3--2--1-+     O: r_monk
|             O  X | O | O  O  O  O       |     0 points
|             O    |   | O  O  O  O       |    
|             O    |   |          O       |    
|                  |   |                  |    
|                  |   |                  |    
^|                  |BAR|                  |     5 point match
|                  |   |                  |    
|                  |   |                  |    
|                  |   | X                |    
| O                |   | X     X  X  X  X |    
| O  X  X     X    |   | X     X  X  X  X |     3 points
+13-14-15-16-17-18------19-20-21-22-23-24-+     X: You (Cube: 2)


I count 14 rolls that make the closeout.  That's apparently worth more than the one extra return shot and the strategic superiority of owning the 5 instead of the three.

I counted return shots, but neglected to count covering shots for us.  Subtle, perhaps, but important, although I doubt I'd ever see it over the board.  I suppose that's the difference between great players and hacks like me.

Hat tip to Nigel Merrigan for the explanation.

[close]

[diane]

Ahhh...that makes sense...thanks!!

[ah_clem]

Ahhh...that makes sense...thanks!!

gnu 2-ply rollout

Spoiler

Two ply rollout doesn't change the story.  Making the three is correct.

One other reason that making the three might be better is that we're just trying to win and if we're going to have an opposing checker hanging around in our homeboard, it's better to have it on the five than the three.  Sometimes when you just want to transition to a race it's actually better to make the deep homeboard points. 



    1. Rollout          9/3 4/3                      Eq.:  +0.412
       0.759 0.422 0.002 - 0.241 0.071 0.005 CL  +0.412 CF  +0.412
      [0.001 0.011 0.001 - 0.001 0.001 0.000 CL   0.003 CF   0.003]

    2. Rollout          11/5 6/5                     Eq.:  +0.382 ( -0.031)
       0.751 0.389 0.004 - 0.249 0.080 0.006 CL  +0.382 CF  +0.382
      [0.001 0.010 0.001 - 0.001 0.001 0.001 CL   0.003 CF   0.003]
        Full cubeful rollout with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 900864643 and quasi-random dice
        Play:  2-ply cubeful prune [world class]
        keep the first 0 0-ply moves and up to 8 more moves within equity 0.12
        Skip pruning for 1-ply moves.
        Cube: 2-ply cubeful prune [world class]

[close]

[PersianLord]

Well, a great position indeed. Thank you Diane.

I think that the main consideration here, as other fellow fibsters pointed out, is the 'building opportunities'. After making the 5-pt, our resources to build the 3-pt and close-out will be depleted as the checker on 10-pt is 7 point away from the 3-pt and thus, 'out of building range'; while if we opt to make the 3-pt first, this won't be the case as all of our men will be in building range.

About the 'strategical' importance of the 5-pt, I think we should not exaggerate significance of the so-called 'golden point' too much. Yes, it's good to make your 5-pt, but 'early' in the game and not during a blitz or pre-close-out period, as during these last stages having a 'prime' is of least value. During blitz/close-out, the most important thing to consider is to not allowing your opponent to make an anchor, especially in your 'deep' points. And if by any means, we're forced to let our opponent making an anchor, let him make an 'advanced' anchor as you'll have less hassle bearing-off your men against an advanced anchor than when facing a deep anchor. Correct me if I'm wrong.

Just my 2 cents...

PL

[ah_clem]

gnu 2-ply rollout

Spoiler

in case there are still lingering doubts.


    1. Rollout          9/3 4/3                      Eq.:  +0.412
       0.759 0.422 0.002 - 0.241 0.071 0.005 CL  +0.412 CF  +0.412
      [0.001 0.011 0.001 - 0.001 0.001 0.000 CL   0.003 CF   0.003]

    2. Rollout          11/5 6/5                     Eq.:  +0.382 ( -0.031)
       0.751 0.389 0.004 - 0.249 0.080 0.006 CL  +0.382 CF  +0.382
      [0.001 0.010 0.001 - 0.001 0.001 0.001 CL   0.003 CF   0.003]
        Full cubeful rollout with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 900864643 and quasi-random dice
        Play:  2-ply cubeful prune [world class]
        keep the first 0 0-ply moves and up to 8 more moves within equity 0.12
        Skip pruning for 1-ply moves.
        Cube: 2-ply cubeful prune [world class]

[close]