Match 1, Game 3, Move 12, Forum

[Zorba]

I'll fill in for blitzxz while he's away for the weekend. Never mind the different board  

Match to 5, factotum 0, FIBSBoard forum 2

Pipcount: factotum 144 FIBSBoard forum 119

GnuBG ID: cLfBBAN2nA0MIA:UomxAAAAEAAA

FIBSBoard forum to play 4-3:

[diane]

Spoiler

I cant imagine doing anything other than getting up to that prime with a view to escaping, so 24-21 is definite.  Then since Zorba has removed the possibility of me picking 15-11 as the other half of that move, I wont do that      The only real thing left for the four to me is the 6-2, much as once again, it isn't something I enjoy doing..

[close]

[Zorba]

Spoiler

Stepping up is a must and 6/2 for safety. Next! 

[close]

[Zorba]

I'm sorry I totally missed the brilliant mass-duplication of 15/11 

[ah_clem]

15/11 for sure.  We can explain the "error" to blitzyz when he gets back.

[Zorba]

Ok, I'll "next" this one, hard to see any other move winning the vote...

[playBunny]

GnuBg 3-ply gives a -0.026 equity difference for the second move, 24/21 8/4 but the third move is definitely a blunder. 8/4 is perhaps something of a bot move but someone might have seen it, like it and argued for it. (Not me, though; I didn't see it, lol). Perhaps it's better for the boardmaster to check Gnubg before automatically nexting?

[Zorba]

Yes, well I thought about nexting it right away, then noticed GnuBG had 8/4 quite close, so I put up the vote just in case. With 5 votes all for 6/2 I thought I may as well continue from here to a more interesting play/vote. It's a judgement call to determine what exactly is a "next" but I thought this one qualified, as it's hard to see anyone preferring 8/4 over 6/2 here, even when 8/4 is not much worse equity-wise. But maybe people rather have the full 2 day voting period for plays like this? Opinions welcome!

[playBunny]

Ah, I see. There aren't any votes shown next to the options so I wasn't aware that the move-o-meter had already swung that much.

[Zorba]

Thanks for the feedback, I thought the vote count would show up after I locked the poll, but after logging in as "guest" I see it's not there.  That's confusing, so I guess it's better to keep the voting open.

Is everybody okay with the 6/2 play and continuing the match? I think I might have been a little too impatient here given the rollout result below, sorry about that I gues I'm just too curious about the follow-up here

GnuBG 2-ply ROLLOUT:

Spoiler

It's very close, but the obvious 6/2 is still best. I find it a remarkable result, I wouldn't have expected this at all, thinking more of 8/4 as some 0.05-0.10 error and dissing it in a split second. I've been thinking about the play to come up with reasons why it's so close, and may have found a few:

• duplication of hitting threes: if factotum hits the eightpoint blot and not our backman, we may anchor on the fourpoint. It may not improve our winning chances per se, but at least it would reduce our gammon losses
• starting our fourpoint: obviously can't hurt to try and make that point, we have enough gaps already
• luring factotum off his anchor: if factotum breaks it to hit that blot, we're not happy but we do get a new winning plan. We may now succeed in attacking factotum's backmen, and if we get him to dance on the bar and improve our board further that may help us escape and get home safely. So, the blot on eight is a target for factotum, but not as good as it may seem at first
• stripped position: after 6/2 we are so stripped and inflexible, that we may leave a blot anyhow, and it might be more inconvenient then

        Full cubeless rollout (trunc. at one-sided bearoff) with var.redn.
        1296 games, Mersenne Twister dice gen. with seed 859211562 and quasi-random dice
        Play:  2-ply cubeless prune
        keep the first 0 0-ply moves and up to 8 more moves within equity 0,08
        Skip pruning for 1-ply moves.

    1. Rollout          24/21 6/2                    Eq.:  -0,4198
        35,76  14,27   1,63 -  64,24  20,33   3,41 CL  -0,4198
      [  0,19   0,36   0,27 -   0,19   0,25   0,36 CL   0,0041]
    2. Rollout          24/21 8/4                    Eq.:  -0,4348 ( -0,0150)
        34,94  15,37   2,39 -  65,06  20,10   2,78 CL  -0,4348
      [  0,19   0,36   0,31 -   0,19   0,25   0,30 CL   0,0042]


And for the record, a quick 0/2-ply rollout of all moves (108-1296 trials):


    1. Rollout          24/21 6/2                    Eq.:  -0,3925
    2. Rollout          24/21 8/4                    Eq.:  -0,4225 ( -0,0300)
    3. Rollout          9/2                          Eq.:  -0,4792 ( -0,0867)
    4. Rollout          8/1                          Eq.:  -0,5140 ( -0,1215)
    5. Rollout          8/4 6/3                      Eq.:  -0,5435 ( -0,1510)
    6. Rollout          15/8                         Eq.:  -0,5546 ( -0,1622)
    7. Rollout          9/6 8/4                      Eq.:  -0,5654 ( -0,1729)

[close]

[playBunny]

I suspect that you're right in thinking that nobody would go for the 8/4 (two more votes for the 6/2, including mine), and my peek at GnuBg advice was already part of your method anyway so you have my vote of confidence, backdated to before my first post.

Spoiler

I agree with your ideas, especially the 4th. The 6/2 declaws that checker and, as we're in a kind of ongoing pay now/pay later situation and there's got to be risk taken at some stage, a risk now, with a home point potential (your 2nd idea), may lessen the risk later (or rather increase our threat).

[close]