Match 1, Game 3, Move 7, Forum

[blitzxz]

5-4?

Score:
Green(forum) 2, White(factotum) 0, match to 5

Pips:
Green 132 White 169

GnuID:
sG3wABOM55gZAA:QQm2AAAAEAAA

[ah_clem]

Spoiler

I'd be cautious about hitting loose  at this point - he's got a two point board, an advanced anchor and a four-prime.  Plus we're ahead in the match and the race.  Exchanging hits is not good for us. However, we have to leave a blot somewhere...so where is best to leave it?  Well, if we hit, we take away half the roll, and a blot on the two is harder to hit than a blot elsewhere.  OTOH, leaving the blot where he has to abandon the anchor has it's charms...and if we can clear a back point whilst doing so we're moving forward.

So 15-10 15-11 Clear the 15 point and leave a tempting blot to lure him off of his anchor. I think this is better than clearing the 13 point since 1) the 13 point will be easier to clear and 2) the stack on the eight that ensues.

No, I'm not sure about this one either...

[close]

[diane]

Spoiler

Same move as ah_clem for me, for pretty much the same reasons, didnt like the loose hit - but do want to start bringing those back ones in, and if they are going to get hit - then lets get that over with now, while he has a two point board.  The split of the anchor, if he hits from there, may work for us too.  [oops  he said that, didnt he ]
I am also very open to persuasion on this one though, I am by no means convinced I am right.

[close]

[lewscannon]

Spoiler

Not thrilled with this roll. no way to avoid leaving a blot. For the same reasons as listed, i'd move the 15s, simply as this seems like the clearest path to victory in this game, if we don't get hit.

[close]

[stog]

Spoiler

11,2 hit, as i would favour keeping the 15 point intact, although this could leave us with a similiar fate next roll, if he fails to hit this roll. if he hits 2 we have better position and timing than if he hits us at 10, although there as u say he might very well have to split his anchor

[close]

[NIHILIST]

Spoiler

15-6. We have a huge race lead, generally the checkers furthest from home are the most difficult to clear. 1 less hitting number than moving both from the 15 point.

It doesn't look like it'll be any easier to clear the 15 point next roll if we make another play, let's do it now.

[close]

Bob

[ah_clem]

Spoiler

15-6. We have a huge race lead, generally the checkers furthest from home are the most difficult to clear. 1 less hitting number than moving both from the 15 point.

There may be less hitting numbers by  leaving the blot on the 15, but he can hit that blot with impunity - there are no adverse consequences of hitting it loose, so he can use the other half of the roll to do something useful.  Or just keep moving.  Conversely, abandoning the anchor to hit on the 10 pt leaves us lots of targets and gives up one of his best assets.  IOW, hitting a blot on the 10pt weakens his position; he can hit on the 15 and maintain his assets.  It seems clear to me that moving both 15 pt checkers is the better move.

[close]

[Zorba]

Spoiler

Hmm, more tough play. Leaving a blot 6 away from opponent's checkers (15/10 15/11) is not the first thing that comes to mind when you don't want to get hit. And with factotum's 4-prime in place as reception committee forr when we arrive on the bar, we sure rather not get hit. But we have to leave a shot somehow. Sometimes, when you have to leave a shot anyway, you might as well put opponent on the bar and "get something for your money". Here, however, that does not seem to be very beneficial to us, it cramps our position to go deep inside our homeboard and we'd still be stuck with several points to clear and less time to do it. With factotum's advanced anchor there's also less to gain from loose hits.

So I prefer the play that, when missed, actually gets us a lot closer to bringing this home.

Between 15/6 and 15/10 15/11, there's less shots after the first, but the hits are more powerful: factotum doesn't have to give up part of his defense to hit and the hit brings in a builder as well, increasing factotum's containment chances. If factotum wants to hit the blot on our 10pt, he might be struggling next roll to clean up his mess, which would make an escape easier for us.

So, 15/10 15/11 for me. At least 4-2 is blocked!

[close]

[ah_clem]

So, the vote is 4 to 4.  What do we do now?

[NIHILIST]

You and Zorba have persuaded me, let's go with your play.

Bob

[blitzxz]

Votes are even so I flipped a coin. The hitting play won.

[playBunny]

That's not correct. Bob changed his vote from 13/6 to the 15s. I would have changed my vote if I hadn't been too late. I went for 11/2* but the 8/3 version is worse.

[ah_clem]

Votes are even so I flipped a coin. The hitting play won.

Wow.  This really is a game of chance.  (c:

[blitzxz]

That's not correct. Bob changed his vote from 13/6 to the 15s. I would have changed my vote if I hadn't been too late. I went for 11/2* but the 8/3 version is worse.

When the voting time ended the vote was even. If somebody have better ideas for solving ties I'm open for ideas.

[playBunny]

Might I suggest that we change the strategy for dealing with even votes from a coin toss to something such as further comments in the thread for a given period?

[diane]

Nihi did state that he was changing his vote, but didn't actually do it in the poll - I expect he hasn't spotted that option yet.  I didn't have time this morning to figure out if I could do it for him, given that this is my working weekend. 

Certainly I would prefer that we talked over tied moves before opting to flip a coin in future matches.

[ah_clem]

Well, there were seven votes to hit and 5 votes to play it safe, so this seems the right outcome even though it's not what I voted for.  So  I would be fine if blitzxz made an executive decision that hitting was the popular choice and made that move.  This is not quite a chouette, but it's close enough and he's the captain.

Granted, the hit/no-hit votes  were each spread across two different moves, but we're going to see that alot - several options that are more or less the same idea that splits the vote.  So, I'd suggest that participants who favor a theme change their votes to the popular choice of that theme - in this case people who want to play it safe should pile up on 15/9 15/10 even if 15/6 is their first choice. 

Alternatively, if there's a tie when the voting closes, extend the voting period another 24 hours and allow folks to change their vote.  Or not, and go with the move that's more in keeping with the theme that got the most votes.

Finally, here's the rollout:

Spoiler

Hitting is a big blunder.  Surprisingly, clearing the 13 point (which nobody voted for) is best, followed closely by clearing the 15.  15/6 is a bigger blunder than hitting, but not as big as the hitting play that won the vote.

Note to Zorba: sorry that I haven't implemented your suggestions for rollout parameters - I'm a hack at doing rollouts, but I'm interested in learning more.  Look for another thread asking about rollout settings in gnu - we can discuss it there without cluttering the forum match thread.

1. Rollout          13/9 13/8                    MWC:  50.93%
       0.584 0.123 0.004 - 0.416 0.085 0.002 CL  55.55% CF  50.93%
      [0.001 0.001 0.000 - 0.001 0.001 0.000 CL   0.07% CF   0.08%]
        Truncated cubeful rollout (depth 11) with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 858042048 and quasi-random dice
        Play: 0-ply cubeful prune [expert]
        Cube: 0-ply cubeful prune [expert]
    2. Rollout          15/11 15/10                  MWC:  50.52% ( -0.41%)
       0.579 0.129 0.004 - 0.421 0.080 0.002 CL  55.57% CF  50.52%
      [0.001 0.001 0.000 - 0.001 0.001 0.000 CL   0.08% CF   0.10%]
        Truncated cubeful rollout (depth 11) with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 858042048 and quasi-random dice
        Play: 0-ply cubeful prune [expert]
        Cube: 0-ply cubeful prune [expert]
    3. Rollout          11/2*                        MWC:  48.17% ( -2.77%)
       0.541 0.126 0.004 - 0.459 0.109 0.003 CL  53.02% CF  48.17%
      [0.001 0.001 0.000 - 0.001 0.001 0.000 CL   0.08% CF   0.08%]
        Truncated cubeful rollout (depth 11) with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 858042048 and quasi-random dice
        Play: 0-ply cubeful prune [expert]
        Cube: 0-ply cubeful prune [expert]
    4. Rollout          13/8 6/2*                    MWC:  48.01% ( -2.93%)
       0.543 0.115 0.003 - 0.457 0.102 0.003 CL  53.00% CF  48.01%
      [0.001 0.001 0.000 - 0.001 0.001 0.000 CL   0.09% CF   0.09%]
        Truncated cubeful rollout (depth 11) with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 858042048 and quasi-random dice
        Play: 0-ply cubeful prune [expert]
        Cube: 0-ply cubeful prune [expert]
    5. Rollout          15/6                         MWC:  47.96% ( -2.97%)
       0.541 0.132 0.004 - 0.459 0.096 0.003 CL  53.45% CF  47.96%
      [0.001 0.001 0.000 - 0.001 0.001 0.000 CL   0.08% CF   0.10%]
        Truncated cubeful rollout (depth 11) with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 858042048 and quasi-random dice
        Play: 0-ply cubeful prune [expert]
        Cube: 0-ply cubeful prune [expert]
    6. Rollout          8/3 6/2*                     MWC:  47.77% ( -3.16%)
       0.536 0.125 0.004 - 0.464 0.115 0.004 CL  52.66% CF  47.77%
      [0.001 0.001 0.000 - 0.001 0.001 0.000 CL   0.08% CF   0.08%]
        Truncated cubeful rollout (depth 11) with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 858042048 and quasi-random dice
        Play: 0-ply cubeful prune [expert]
        Cube: 0-ply cubeful prune [expert]

[close]

[blitzxz]

Alternatively, if there's a tie when the voting closes, extend the voting period another 24 hours and allow folks to change their vote.

We do this then since people want more time. A day's time to change votes/new voter to come. If it's still even then I'll flip a coin.

[Zorba]

Here's my rollout in EMG... we got the double whopper...yay

Spoiler

    1. Rollout          13/9 13/8                    Eq.:  -0,1564
        58,31  20,44   1,75 -  41,69   8,56   0,75 CL  +0,1484 CF  -0,1564
      [  0,13   0,30   0,14 -   0,13   0,10   0,07 CL   0,0033 CF   0,0057]
        Full cubeful rollout with var.redn.
        3888 games, Mersenne Twister dice gen. with seed 857820087 and quasi-random dice
        Play: 0-ply cubeful prune [expert]
        Cube: 0-ply cubeful prune [expert]
    2. Rollout          15/11 15/10                  Eq.:  -0,1715 ( -0,0152)
        58,07  19,68   1,80 -  41,93   7,97   0,77 CL  +0,1524 CF  -0,1715
      [  0,14   0,28   0,14 -   0,14   0,10   0,09 CL   0,0035 CF   0,0063]
        Full cubeful rollout with var.redn.
        3888 games, Mersenne Twister dice gen. with seed 857820087 and quasi-random dice
        Play: 0-ply cubeful prune [expert]
        Cube: 0-ply cubeful prune [expert]
    3. Rollout          13/8 6/2*                    Eq.:  -0,2931 ( -0,1368)
        55,06  17,49   0,72 -  44,94  10,60   0,79 CL  +0,0469 CF  -0,2931
      [  0,41   0,87   0,17 -   0,41   0,34   0,29 CL   0,0111 CF   0,0190]
        Full cubeful rollout with var.redn.
        324 games, Mersenne Twister dice gen. with seed 857820087 and quasi-random dice
        Play: 0-ply cubeful prune [expert]
        Cube: 0-ply cubeful prune [expert]
    4. Rollout          15/6                         Eq.:  -0,3001 ( -0,1437)
        54,72  19,74   1,48 -  45,28   9,09   0,85 CL  +0,0756 CF  -0,3001
      [  0,42   1,04   0,39 -   0,42   0,34   0,26 CL   0,0100 CF   0,0194]
        Full cubeful rollout with var.redn.
        324 games, Mersenne Twister dice gen. with seed 857820087 and quasi-random dice
        Play: 0-ply cubeful prune [expert]
        Cube: 0-ply cubeful prune [expert]
    5. Rollout          11/2*                        Eq.:  -0,3187 ( -0,1623)
        54,04  22,06   1,24 -  45,96  10,31   1,13 CL  +0,0386 CF  -0,3187
      [  0,42   1,35   0,29 -   0,42   0,39   0,35 CL   0,0113 CF   0,0200]
        Full cubeful rollout with var.redn.
        324 games, Mersenne Twister dice gen. with seed 857820087 and quasi-random dice
        Play: 0-ply cubeful prune [expert]
        Cube: 0-ply cubeful prune [expert]
    6. Rollout          15/11 8/3                    Eq.:  -0,3421 ( -0,1857)
        53,84  20,46   2,21 -  46,16  10,00   0,91 CL  +0,0396 CF  -0,3421
      [  0,48   1,10   0,56 -   0,48   0,34   0,32 CL   0,0121 CF   0,0227]
        Full cubeful rollout with var.redn.
        324 games, Mersenne Twister dice gen. with seed 857820087 and quasi-random dice
        Play: 0-ply cubeful prune [expert]
        Cube: 0-ply cubeful prune [expert]
    7. Rollout          8/3 6/2*                     Eq.:  -0,3608 ( -0,2044)
        53,69  20,47   1,48 -  46,31  11,51   1,14 CL  +0,0189 CF  -0,3608
      [  0,43   1,14   0,47 -   0,43   0,36   0,33 CL   0,0113 CF   0,0199]
        Full cubeful rollout with var.redn.
        324 games, Mersenne Twister dice gen. with seed 857820087 and quasi-random dice
        Play: 0-ply cubeful prune [expert]
        Cube: 0-ply cubeful prune [expert]

[close]