Move 105 - End game

[sixty_something]

If the board image does not appear in the question above, you may view it in the attached file below.

one vote per user .. results of the poll are shown after you have voted

then, open the text file below for a copy of the GNUbg analysis or you may enter the ID's below into GNUbg or Snowie to evaluate your answer:

Position ID: AACAsW03KAAAAA  Match ID: Qom1ABAAAAAA

A nation which does not remember what it was yesterday, does not know what it is today, nor what it is trying to do. We are trying to do a futile thing if we do not know where we came from or what we have been about.
-- Woodrow Wilson 28th president of US (1856 - 1924)

[don]

Interesting, particularly with the votes.  This is a position where you need to identify your objective.  The best move is not listed, and there's half a dozen better moves than the worst move listed.

Spoiler

Here your best chance is to get your opponent to leave his anchor on your 1 point, with the intention of getting both his checkers on the bar.  Open up your bar point and hope he rolls a 6, giving you lots of return shots at his blot on your 1 and in your outfield.

Best move:  12/9 7/2 (not listed)

[close]

--
don

[PersianLord]

Interesting, particularly with the votes.  This is a position where you need to identify your objective.  The best move is not listed, and there's half a dozen better moves than the worst move listed.

Spoiler

Here your best chance is to get your opponent to leave his anchor on your 1 point, with the intention of getting both his checkers on the bar.  Open up your bar point and hope he rolls a 6, giving you lots of return shots at his blot on your 1 and in your outfield.

Best move:  12/9 7/2 (not listed)

[close]

--
don

Spoiler

I think that 7/2-6/3 is better, as it adds more assets. If red gets a lucky roll of 6-6 and you have moved 12/9, you will have just 1 hunter blot, the hit one in 7 pt, to attack red's runners, but with 6/3, red will be forced to hit both checkers of 12pt and 7pt and thus white will has 2 hunters and this increases his chances of hitting. Also with 6/3, you can attack the future blot in 1pt, if red get's a 6 and escape one chacker.

This is my analysis and may be I am wrong, but i will play 7/2-6/3, as i trust GNUBG in almost every move it makes.

[close]

[blitzxz]

Spoiler

This should be some sort of reference position. Recycling checkers is an important thema. When the other player has crushed board (or bear-off board in this case), you lose nothing if you get your checkers send back. But you can win a lot by hitting more checkers or breaking anchor. Recycling is also used to ruin backgame players timing. I made a wrong quick vote here and forgot the whole thema.

[close]

[spielberg]

A small point but an important one I feel PersianLord - please try and use the "spoiler" button to avoid giving the answer in plain text for some to read.

Spoiler

I was SO proud to have actually got this one right too - tho' to be honest I suspected it was a very unusual one so I thought "outside the box".

[close]

[sixty_something]

The best move is not listed, and there's half a dozen better moves than the worst move listed.

that's why i added Other as an option .. change your vote to Other, amigo, if you think a better move is not listed .. sure is a pretty little Buddha you got there, too .. can't wait to see the pics of him from his worldwide travels

No law or ordinance is mightier than understanding.
-- Plato , Laws Greek author & philosopher in Athens (427 BC - 347 BC)

[dorbel]

An interesting problem but utterly facile analysis. Gnu 0-ply evaluation isn't worth tuppence and to assume from it a "right" answer is absurd.
How long does it take to do a rollout?

[blitzxz]

Here's the rollout (2-ply, narrow filter). But it's not enough to call the best move. Differences are very small.

Spoiler

    1. Rollout          12/9 7/2                     Eq.:  -0,052
       0,474 0,000 0,000 - 0,526 0,051 0,034 CL  -0,052 CF  -0,052
      [0,003 0,000 0,000 - 0,003 0,003 0,004 CL   0,006 CF   0,006]
        Full cubeful rollout with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 1055124781 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]
    2. Rollout          13/8 7/4                     Eq.:  -0,056 ( -0,004)
       0,472 0,000 0,000 - 0,528 0,072 0,050 CL  -0,056 CF  -0,056
      [0,004 0,000 0,000 - 0,004 0,004 0,005 CL   0,007 CF   0,007]
        Full cubeful rollout with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 1055124781 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]
    3. Rollout          13/10 7/2                    Eq.:  -0,060 ( -0,008)
       0,470 0,000 0,000 - 0,530 0,072 0,047 CL  -0,060 CF  -0,060
      [0,003 0,000 0,000 - 0,003 0,003 0,003 CL   0,006 CF   0,006]
        Full cubeful rollout with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 1055124781 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]
    4. Rollout          7/2 6/3                      Eq.:  -0,077 ( -0,025)
       0,462 0,000 0,000 - 0,538 0,054 0,042 CL  -0,077 CF  -0,077
      [0,002 0,000 0,000 - 0,002 0,004 0,004 CL   0,005 CF   0,005]
        Full cubeful rollout with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 1055124781 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]
    5. Rollout          7/4 7/2                      Eq.:  -0,085 ( -0,033)
       0,458 0,000 0,000 - 0,542 0,039 0,024 CL  -0,085 CF  -0,085
      [0,002 0,000 0,000 - 0,002 0,002 0,002 CL   0,005 CF   0,005]
        Full cubeful rollout with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 1055124781 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]

[sixty_something]

An interesting problem but utterly facile analysis. Gnu 0-ply evaluation isn't worth tuppence and to assume from it a "right" answer is absurd.
How long does it take to do a rollout?

thanks for the rollouts, blitzz .. i'm still laughing at "utterly facile analysis" .. regardless, it's a good point, dorbel, and expressed in an utterly dorbelian manner  i had to look up "facile" to understand your point

facile
Pronunciation: \ˈfa-səl\
Function: adjective
Etymology: Middle French, from Latin facilis, from facere to do â€" more at do
Date: 15th century
1 a (1): easily accomplished or attained <a facile victory> (2): shallow, simplistic
b: used or comprehended with ease c: readily manifested and often lacking sincerity or depth <facile tears>
2 archaic : mild or pleasing in manner or disposition
3 a: ready, fluent <facile prose> b: poised, assured

nobody, including myself, anointed me problem editor and i am certainly no expert .. the intent of these problems, for me, remains allowing anyone to propose a problem and see how others answer it .. hopefully giving experts and advanced GNUbg users and hacks, like me, an opportunity to provide elaboration beyond basic analysis, when warranted

so, dorbel, in all seriousness, care to offer some critical commentary on the problem and the manner in which it should be analyzed more arduously with GUNbg? because, frankly, it's not yet clear to me what the best analysis approach here would be

Spoiler

for example, when i did these roll-outs, for the 0-ply recommended move 7/2 6/3 i got results that varied substantially from the 0-ply evaluation (Eq.:  -0.087) .. the roll-outs at 2-ply [world class] ran substantially longer and produced results that are still inconclusive regarding selecting a "best move" .. regardless, i think it has turned into an interesting problem and am eager to see what a more thourough analysis may yield

    1. Rollout          7/2 6/3                      Eq.:  -0.139
       0.431 0.000 0.000 - 0.569 0.052 0.022 CL  -0.139 CF  -0.139
      [0.003 0.000 0.000 - 0.003 0.001 0.001 CL   0.006 CF   0.006]
        Truncated cubeful rollout (depth 11) with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 1055060082 and quasi-random dice
        Play: 0-ply cubeful prune [expert]
        Cube: 0-ply cubeful prune [expert]
~~~~~~~~~
    1. Rollout          7/2 6/3                      Eq.:  -0.142
       0.429 0.000 0.000 - 0.571 0.040 0.016 CL  -0.142 CF  -0.142
      [0.002 0.000 0.000 - 0.002 0.002 0.001 CL   0.004 CF   0.004]
        Truncated cubeful rollout (depth 11) with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 1055060082 and quasi-random dice
        Play:  2-ply cubeful prune [world class]
        keep the first 2 0-ply moves and up to 3 more moves within equity 0.1
        keep the first 0 1-ply moves
        Cube: 2-ply cubeful prune [world class]

~~~~~~~~~~

while this move (13/8 7/4) was not one included as an optional answer, it produced a lower equity loss at 2-ply than did the 7/2 6/3 above and illustrates don's comment about "other" moves perhaps being better

    1. Rollout          13/8 7/4                     Eq.:  -0.136
       0.432 0.000 0.000 - 0.568 0.045 0.018 CL  -0.136 CF  -0.136
      [0.002 0.000 0.000 - 0.002 0.002 0.002 CL   0.004 CF   0.004]
        Truncated cubeful rollout (depth 11) with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 1055060082 and quasi-random dice
        Play:  2-ply cubeful prune [world class]
        keep the first 2 0-ply moves and up to 3 more moves within equity 0.1
        keep the first 0 1-ply moves
        Cube: 2-ply cubeful prune [world class]

rather than randomly try others, i'll await someone describing how to use GNUbg rollouts to perform the in depth analysis this little problem apparently deserves (actually, i'm afraid being labelled a "facile" analyst )

for practical purposes and dare i say facile play, this position may be an example where analysis produces no clear readily seen tactical solution .. the strategic picture of how to approach the position however has two clear alternatives .. the approach of breaking the prime in an attempt to get two back and on the bar for a greater chance at winning is one i had not seen before .. thus, imo, it's a worthwhile little problem

[close]

I am not concerned…with offering any facile solution for so complex a problem
â€" T. S. Eliot, OM (1888â€"1965), American born British subject, poet, dramatist, and literary critic. He received the Nobel Prize in Literature in 1948. 

[dorbel]

Spoiler

Making one of the plays that forces the defender to hit with an immediate 6 is an imaginative play, but it isn't clearly correct. Rollouts of this position show that just clearing the bar point seems to do almost as well. Either of these strategies are however superior to just meekly bringing checkers home. Trapping Red off the anchor before the bearoff starts has to be good.
This the sort of position that you can learn a lot from if you roll it out yourself. Note that White has a difficult play here and will get many more before the end. Red's play is going to be simplicity itself. If Gnu tells you that White has a certain equity after any play here, you may assume that you are going to have to play brilliantly in order to achieve it. If you make Red hit with the next 6, I think that you are going to make your future play even more difficult, whether he hits or not. For example, if he doesn't roll a 6, are you going to leave the blot there next roll. Not sure? See what I mean? I vote for 7/4, 7/2 to keep it simple.

[close]

[blitzxz]

Spoiler

I tried dorpel's advice and played it couple of times to the end. The moves for white can truly be hard ones. For example after 12/9 7/2 and no six for red if now white rolls 4-1 it is correct to leave even more shots with 9/5 4/3. The same leaving blots thema continues also later unless white manages to close the board or can bear-off and there is no hits. Also after 7/4 7/2 gnu likes to leave blots in many positions later.

[close]